static-curis.ku.dk · availability is rather high ( Vorholt 2012; Carvalho and Castillo 2018)....
Transcript of static-curis.ku.dk · availability is rather high ( Vorholt 2012; Carvalho and Castillo 2018)....
u n i ve r s i t y o f co pe n h ag e n
Genomics of Aerobic Photoheterotrophs in Wheat Phyllosphere Reveals DivergentEvolutionary Patterns of Photosynthetic Genes in Methylobacterium spp
Zervas Athanasios Zeng Yonghui Madsen Anne Mette Hansen Lars H
Published inGenome Biology and Evolution
DOI101093gbeevz204
Publication date2019
Document versionPublishers PDF also known as Version of record
Document licenseCC BY
Citation for published version (APA)Zervas A Zeng Y Madsen A M amp Hansen L H (2019) Genomics of Aerobic Photoheterotrophs in WheatPhyllosphere Reveals Divergent Evolutionary Patterns of Photosynthetic Genes in Methylobacterium sppGenome Biology and Evolution 11(10) 2895-2908 httpsdoiorg101093gbeevz204
Download date 22 jun 2020
Genomics of Aerobic Photoheterotrophs in Wheat
Phyllosphere Reveals Divergent Evolutionary Patterns of
Photosynthetic Genes in Methylobacterium spp
Athanasios Zervas 1 Yonghui Zeng12 Anne Mette Madsen3 and Lars H Hansen141Section of Environmental Microbiology and Biotechnology Department of Environmental Science Aarhus University Roskilde Denmark2Aarhus Institute of Advanced Studies Aarhus University Denmark3The National Research Centre for the Working Environment Copenhagen Denmark4Environmental Microbial Genomics Group Department of Plant and Environmental Sciences University of Copenhagen Frederiksberg Denmark
Corresponding authors E-mails zengenvsaudk lhhaplenkudk
Accepted September 11 2019
Data deposition This project has been deposited at GenBank under the accession numbers SAMN12513103ndashSAMN12513123
Abstract
Phyllosphere is a habitat to a variety of viruses bacteria fungi and other microorganisms which play a fundamental role in
maintaining the health of plants and mediating the interaction between plants and ambient environments A recent addition to
this catalogueofmicrobialdiversitywas theaerobicanoxygenicphototrophs (AAPs) agroupofwidespreadbacteria thatabsorb light
through bacteriochlorophyll a (BChl a) to produce energy without fixing carbon or producing molecular oxygen However culture
representatives of AAPs from phyllosphere and their genome information are lacking limiting our capability to assess their potential
ecological roles in this unique niche In this study we investigated the presence of AAPs in the phyllosphere of a winter wheat
(Triticum aestivum L) in Denmark by employing bacterial colony based infrared imaging and MALDI-TOF mass spectrometry (MS)
techniques A total of 4480 colonies were screened for the presence of cellular BChl a resulting in 129 AAP isolates that were
further clustered into 21 groups based on MALDI-TOF MS profiling representatives of which were sequenced using the Illumina
NextSeq and Oxford Nanopore MinION platforms Seventeen draft and four complete genomes of AAPs were assembled belonging
in Methylobacterium Rhizobium Roseomonas and a novel Alsobacter We observed a diverging pattern in the evolutionary rates of
photosynthesis genes among the highly homogenous AAP strains of Methylobacterium (Alphaproteobacteria) highlighting an
ongoing genomic innovation at the gene cluster level
Key words AAP bacteria bacteriochlorophyll phyllosphere MALDI-TOF MS Nanopore photosystems
Introduction
Plantndashmicrobe interactions both above (phyllosphere) and
below (rhizosphere) ground are common in nature
Traditionally these relationships are investigated in the rhizo-
sphere where conditions are relatively stable and nutrient
availability is rather high (Vorholt 2012 Carvalho and
Castillo 2018) Despite recent work (reviewed in eg
Lindow and Brandl 2003 Vorholt 2012) the microbendash
phyllosphere (which is comprised by the aerial partsmdashand
especially the leavesmdashof plants) interactions remain relatively
understudied compared with rhizosphere Models suggest
that the total leaf surface is gt1 billion km2 (Woodward and
Lomas 2004) potentially colonized by up to 1026 bacterial
cells (Lindow and Brandl 2003)
Although the phyllosphere seems to be a common envi-
ronment for bacteria to thrive in it may be ldquoextremerdquo for a
plethora of reasons First different plants exhibit different
growth patterns and climate adaptations Annual plants com-
plete their life cycle in just one growth season whereas pe-
rennial plants grow and shed leaves every year This creates a
discontinuous ever-changing habitat (Vorholt 2012)
Meanwhile environmental changes also affect the phyllo-
sphere and its inhabitants Winds rainfall frost and drought
all play important roles in shaping the conditions encountered
The Author(s) 2019 Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (httpcreativecommonsorglicensesby40) which permits unrestricted reuse
distribution and reproduction in any medium provided the original work is properly cited
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2895
GBED
ownloaded from
httpsacademicoupcom
gbearticle-abstract111028955570988 by Royal Library C
openhagen University user on 16 January 2020
by microorganisms of the phyllosphere However the most
significant environmental driver is sunlight Temperature dif-
ferences in the upper leaves especially those that form the
canopy of various habitats may range up to 50 C between
day and night Sunlight though beneficial for photosynthetic
organisms is also comprised of UV-radiation which is dam-
aging to the DNA of both prokaryotes and eukaryotes
(Remus-Emsermann et al 2014) In addition to the abiotic
factors that make the phyllosphere a much more hostile en-
vironment than the rhizosphere its inhabitants also facemdash
directly or indirectlymdashthe influence of their plant host
Leaves are surrounded by a thin laminar and waxy layer
the cuticle which renders the leaf surface hydrophobic
thus removing the excess of water that is collected due to
rainfall dew or respiration from the stomata This leads to the
retention of little water most usually in veins and other cavities
of the cuticle These microformations also protect the colo-
nizers from the surrounding environment (Beatie 2002) Apart
from niche competition and scarce water availability phyllo-
sphere colonizers also need to compete for nutrients as the
cuticle makes the surface virtually impermeable to nutrients
deriving from diffusion through the cells of the plant host
(Wilson and Lindow 1994a 1994b) At the same time they
also need to protect themselves from potential invaders and a
plethora of antimicrobial compounds of prokaryotic or eu-
karyotic origin (Vorholt 2012)
Albeit the harsh conditions they encounter phyllo-
sphere microogransims exhibit remarkable biodiversity
which in certain cases can be comparable to that of hu-
man gut microbiome (Xu and Gordon 2003 Knief et al
2012) Through recent genomics and metagenomics stud-
ies several bacterial genera such as Methylobacterium
have been shown to be ubiquitous in the phyllosphere
(Knief et al 2010) and their role in nutrient recycling plant
growth promotion and protection has been evidenced pre-
viously (Beattie and Lindow 1999 Gourion et al 2006
Atamna-Ismaeel and Finkel 2012a Kwak et al 2014)
One of the most interesting recent findings in phyllosphere
microbiota was the presence of aerobic anoxygenic photo-
trophic (AAP) bacteria and rhodopsin-harboring bacteria in
a variety of land plants (Atamna-Ismaeel and Finkel 2012a
2012b) AAP bacteria are commonly found in aquatic envi-
ronments ranging from the arctic to the tropics and from
high salinity lakes to pristine high altitude lakes (Yurkov and
Hughes 2017) They rely on organic carbon compounds to
cover their nutritional requirements and can also utilize
light to produce energy in the form of ATP without fixing
carbon or producing oxygen AAPs have been found in a
variety of extreme environments and have shown their po-
tential to outgrow nonAAP bacteria under nutrient limiting
conditions (Hauruseu and Koblızek 2012 Ferrera et al
2017) Thus their presence in the phyllosphere may not
be surprising However their relative abundance at least
in the rice phyllosphere (Oryza sativa) was up to three
times higher than what is commonly found in marine eco-
systems (Atamna-Ismaeel and Finkel 2012a)
These early culture-independent metagenomics studies
provided valuable information about the presence of AAP
bacteria in the phyllosphere However no pure cultures of
AAPs have been described so far from phyllosphere and
thus detailed genome information is lacking which prevents
an in-depth understanding of the ecological roles and evolu-
tionary trajectory of AAPs in this unique niche To expand our
genomics views of this ecologically important group of bac-
teria in the phyllosphere we designed this study with the
following aims 1) to create the first collection of AAP isolates
from the phyllosphere 2) to characterize their photosynthesis
gene clusters (PGCs) and compare their gene contents and
molecular evolutionary patterns and 3) to expand the data-
base of complete genomes of AAP bacteria from nonaquatic
environments By combining colony infrared (IR) imaging
MALDI-TOF mass spectrometry (MS) and Illumina and
Oxford Nanopore sequencing technologies we were able to
provide 20 draft genomes that contain complete PGCs and
four complete genomes that contain plasmids The further
comparison of the evolutionary trajectories of their PGCs
revealed a diverging pattern among the highly homogenous
Methylobacterium strains highlighting an ongoing genomic
innovation at the gene cluster level
Materials and Methods
Sample Collection
A whole intact plant of winter wheat (Triticum aestivum L)
with a height of circa 60 cm was collected from a field in
Roskilde Denmark on June 6 2018 and transported to the
lab for bacterial isolation on the same day Wheat leaves were
cut off and cleaned with running sterile water to remove dust
and other temporary deposits from the ambient environment
Microbiota were collected from eight pieces of leaves by rins-
ing the leaves in 80 ml PBS solution (pH 74) in a Oslash 150-mm
petri dish and scraping the leave surface repeatedly with ster-
ile swabs until the PBS solution slightly turned greenish due to
the fall-off of leave cells From the supernatant 14 ml were
plated on 15 strength R2A plates (resulting in a total of 137
agar plates) and incubated at room temperature with normal
indoor light conditions for two weeks AAP bacteria were
identified using a custom screening chamber equipped with
an IR imaging system described in detail previously (Zeng
et al 2014) In short plates were placed in a chamber sealed
from light and exposed to a source of green lights AAP bac-
teria containing bacteriochlorophyll absorb the green light
and emit radiation in the near IR spectrum which was cap-
tured by a NIR sensitive CMOS camera The image was proc-
essed and the glowing colonies indicated the presence of
BChl a The colonies that give positive signals were marked
restreaked onto 15 strength R2A plates and incubated at
room temperature for 72 h Verification of the light absorbing
Zervas et al GBE
2896 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
capabilities of the isolated strains was performed using the
same setup Strains that passed the verification step were
further restreaked onto 15 strength R2A plates to ensure
pure colonies of single AAP strains
Selection of AAP Strains and Whole Genome Sequencing
The isolated AAP strains were grouped and dereplicated using
a MALDI-TOF MS (Microflex LT Bruker Daltonics Bremen
Germany) before performing whole genome sequencing to
reduce the cost while maintaining the biodiversity Briefly a
toothpick was used to transfer a small amount of an AAP
bacterial colony onto the target plate (MSP 96 polished steel
Bruker) that was evenly spread out and formed a thin layer of
biomass on the steel plate The sample was then overlaid with
70 formic acid and allowed for air dry before addition of
1lL MALDI-MS matrix solution (a-cyano-4-hydroxycinnamic
acid SigmandashAldrich) A bacterial standard with well-
characterized peaks (Bruker Daltonics) was used to calibrate
the instrument The standard method ldquoMBT_AutoXrdquo was ap-
plied to obtain proteome profiles within the mass range of 2ndash
20 kDa using the flexControl software (Bruker) The
flexAnalysis software (Bruker) was used to smooth the data
subtract baseline and generate main spectra (MSP) followed
by a hierarchical clustering analysis with the MALDI Biotyper
Compass Explorer software which produced a dendrogram
as the output for visual inspection of similarities between
samples An empirical distance value of 50 was used as the
cutoff for defining different groups on the dendrogram cor-
responding to different strainsspecies
From each group on the dendrogram one isolate was cho-
sen and restreaked on 1=2 R2A plates and incubated at room
temperature for one week Prior to DNA isolation the plates
were again tested for the presence of light absorbing pigments
as previously described Bacterial colonies were scraped from
the plates using a loop and immersed in 400mL PBS solution
vortexed until the cells had separated spun down in a table-top
centrifuge at 10000 g at 4 C the supernatant was removed
and the pellets were redissolved in milliQ H2O High molecular
weight DNA was extracted from each strain using a modified
Masterpure Complete DNA amp RNA Purification Kit (Lucigen) by
replacing the elution buffer with a 10mM TrisndashHClmdash50mM
NaCl (pH 75ndash80) solution at the final step DNA concentration
was quantified on a Qubit 20 (Life Technologies) using the
Broad Range DNA Assay kit and DNA quality was measured
on a Nanodrop 2000C (Thermo Scientific) Whole genome
shotgun sequencing was performed on all AAP strains on the
Illumina Nextseq platform in house using the 2150bp chem-
istry Selected strains were also sequenced on the MinION plat-
form (Oxford Nanopore Technologies UK) in house using the
Rapid Barcoding Kit (SBQ-RBK004) and the FLO-MIN106 flow
cell (R941) following the manufacturerrsquos instructional
manuals
Genome Assembly and Analyses
The Illumina pair end reads were trimmed for quality and
ambiguities using Cutadapt (Martin 2011) and assembled
using SPAdes (Bankevich et al 2012) The resulting assem-
blies were analyzed using dRep (Olm et al 2017) to compare
their average nucleotide identities (ANI) in order to identify
clusters of closely related taxa The assemblies were
imported in Geneious R1125 (Biomatters Ltd) and PGCs
were annotated using the Roseobacter litoralis strain
Och149 plasmid pRL0149 (CP002624) as reference under
relaxed similarity criteria (40) The suggested genes of
probable PGCs were analyzed using BLAST (megablast
BlastN) to verify their annotations For full genome annota-
tions the assemblies were uploaded on RAST (Aziz et al
2008 Overbeek et al 2014 Brettin et al 2015) The pre-
dicted translated protein coding genes were imported in
CMG-Biotools (Vesth et al 2013) for pairwise proteome
comparisons using the native blastmatrix program
Hybrid assemblies of pair-end Illumina reads and long
Oxford Nanopore Technologies reads were performed using
the Unicycler assembler (Wick et al 2017) utilizing the bold
assembly mode The resulting complete genomes were
imported in Geneious where their circularity was confirmed
by mapping-to-reference runs using the short and long reads
from the respective hybrid assemblies and the native
Geneious mapper under default settings in the Mediumndash
Low sensitivity mode
Phylogenetic Analyses
Identification of the 16S ribosomal DNA sequences was per-
formed by rnammer in CMG-Biotools Data sets containing
the different genes from the identified PGCs were created in
Geneious The resulting sequences were aligned using MAFFT
v7388 (Katoh et al 2002) and it FFT-NS-I x1000 algorithm
with default settings as implemented in Geneious
Phylogenetic trees were created using RAxML (Stamatakis
2006) employing the GTR-GAMMA nucleotide substitution
model and the rapid bootstrapping and search for best scor-
ing ML-tree with 100 bootstraps replicates and starting from
a random tree Analysis of the PGCs was performed on five
selected genes that were present in all isolates namely acsF
bchL bchY crtB pufL and pufM Individual gene alignments
and phylogenetic trees were performed as described above
A super-matrix (8240 characters) comprised of the total
seven gene alignments was created using the built-in
ldquoConcatenate alignmentsrdquo function in Geneious and a phy-
logenetic tree was created as above All trees were visually
inspected for disagreements Finally a phylogenomic tree
was constructed in RAxML employing the GAMMA
BLOSUM62 substitution model and the rapid bootstrapping
and search for best scoring ML-tree algorithm with 100 boot-
straps replicates and starting from a random tree using as
input the core protein alignment consisting of 6988
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2897
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
characters created with CheckM (Parks et al 2015) and its
lineage_wf algorithm with default settings Substitution rates
were calculated as distances from the root using the legend
provided by the phylogenetic trees
Results
Identification and Isolation of AAP Bacteria from thePhyllosphere
The 129 phyllosphere isolates that were tested positive for the
presence of bacteriochlorophyll-a were placed in 21 groups
according to their protein mass profiles based on MALDI-TOF
MS (supplementary fig MALDI-TOF Supplementary Material
online) A representative from each group was sequenced
(Illumina) and their genomes assembled (SPAdes) WL1 and
WL2 were from the same MALDI-TOF group and their
genomes are identical (ANI 100) confirming the validity
of MALDI-TOF MS for grouping similar isolates Information
about their phylogeny total genome size and genome as-
sembly statistics are presented in table 1
Analysis of the 16S genes showed that 17 isolates
belonged in Methylobacterium (separated in four distinct
groups) whereas the remaining three belonged to
Rhizobium Roseomonas and one novel species was affiliated
to the Methylocystaceae family (WL4) with most hits belong-
ing to uncultured environmental strains The only character-
ized hits of WL4 belonged to Methylosinus trichosporium
(Gorlach et al 1994) and Alsobacter metallidurans a recently
characterized novel genus novel species (Bao et al 2006)
However in both instances the similarity was low
(985) We included strains of both species and created
a 16S phylogenetic tree (fig 1) which placed Alsobacter sp
WL4 as sister to the Alsobacter-Methylosinus clade
The core proteome analysis (fig 2mdashCheckM) including M
trichosporium OB3b and Alsobacter sp SH9 established the
phylogeny of strain WL4 and it was highly supported as sister
to Alsobacter sp SH9 (BSfrac14 100) both of which were
distinct from M trichosporium OB3b (BSfrac14 100)
Methylobacterium strains were still placed into four distinct
groups (Group 1 WL9 WL19 WL69 Group 2 WL1 WL2
WL7 WL18 WL64 Group 3 WL6 WL30 WL93 WL116
WL119 Group 4 WL8 WL12 WL103 WL120) This was
further supported by the whole proteome comparison
(fig 3mdashBLAST matrix) For Groups 3 and 4 the average
core protein content is 4500 proteins whereas for Group
2 it is close to 4000 This may signify the presence of
either more or larger plasmids that characterize the strains
of this group For Group 1 this number falls even more as
in this case there clearly are three distinct species that make
up the group
We expanded the analysis to include 100 complete and
draft Methylobacterium spp genomes available in GenBank
(last assessed April 2019) and also other representatives from
the Methylocystaceae aiming at further establishing strain
WL4 in the genus Alsobacter and also providing greater res-
olution on the Methylobacterium isolates with a special focus
on the divergent group 1 Information about the included
genomes is given in supplementary table 1 Supplementary
Material online We performed whole genome comparisons
using ANI (fig 4mdashdRep)
The analysis showed that Group 1 isolates belonged in
different Methylobacterium species groups 3 and 4 were dif-
ferent strains of the same species (MASH ANI identity be-
tween groupsgt 90) whereas group 3 isolates are closely
related strains possibly belonging to Methylobacterium bra-
chiatum Strain WL4 is again placed close to the two identical
Alsobacter genomes (Alsobacter sp SH9 and A metallidurans
PVZS01) far away from the three remaining species from
within the Methylocystaceae (M trichosporium OB3b
Methylocystis rosea GW6 Methylocystis bryophila S285)
However the MASH-ANI identity between WL4 and
Alsobacter is below the 80 threshold of genus assignment
Therefore we took the top 200 megablast hits for WL4rsquos 16S
region (information provided in supplementary table 2
Supplementary Material online) aligned them including strain
WL4 as mentioned previously and created a phylogenetic tree
of the 201 sequences (supplementary fig 1 Supplementary
Material online) The results clearly show that WL4 belongs in
Alsobacter whereas Methylosinus spp form their own dis-
tinct cluster
Table 1
Genome Statistics and Phylogenetic Information of the 21 Isolates that
were Illumina Sequenced in This Study
Strain Taxonomy Contigs Genome Size N50 GC
WL1 Methylobacterium 654 6258106 19384 6892
WL2 Methylobacterium 298 6292310 52057 6899
WL3 Rhizobium 55 5333961 210875 6114
WL4 Alsobacter 139 5412272 86934 6792
WL6 Methylobacterium 708 5549874 15034 6972
WL7 Methylobacterium 431 6059272 36478 6904
WL8 Methylobacterium 225 5322599 56100 6969
WL9 Methylobacterium 221 4670253 42885 6741
WL12 Methylobacterium 271 5545758 53038 694
WL18 Methylobacterium 1373 5796709 7880 6864
WL19 Methylobacterium 330 5177712 31633 6698
WL30 Methylobacterium 377 5636306 38077 6978
WL45 Roseomonas 2056 5961800 5511 6954
WL64 Methylobacterium 365 6908402 48840 6834
WL69 Methylobacterium 225 4693485 43639 6951
WL93 Methylobacterium 719 5642324 16902 6945
WL103 Methylobacterium 1327 5167285 7154 6897
WL116 Methylobacterium 1173 5445855 8413 6943
WL119 Methylobacterium 263 5709790 45485 6958
WL120 Methylobacterium 395 5266148 26698 6954
WL122 Methylobacterium 2024 4508072 3666 6877
Zervas et al GBE
2898 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Molecular Evolution of the PGC
The PGCs identified in the sequenced isolates all belong to
RC-II type meaning they contain bacteriochlorophyll-a and
possess pheophytinmdashquinone type reaction centers All
strains possess a full bacteriochlorophyll synthase gene set
except for Alsobacter sp WL4 that is lacking the bchM
gene (table 2)
For the carotenoid biosynthesis genes all strains possess
crtB whereas crtA is only present in Alsobacter sp WL4 and
Rhizobium sp strain WL3 For the crtC and crtK genes an
interesting pattern is observed Methylobacterium group 1
and group 2 strains contain crtC (with three exceptions dis-
cussed below) whereas group 3 and 4 strains completely lack
the gene On the other hand crtK is only present in
FIG 1mdashPhylogenetic tree of the 16S genes of the sequenced isolates Methylosinus trichosporium and Alsobacter metallidurans were included to better
pinpoint the phylogenetic placement of strain WL4 The legend shows substitution rates from the root of the tree and the node labels indicate bootstrap
support values ()
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2899
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Methylobacterium group 3 strains and missing from groups 1
2 and 4 Similar patterns are evidenced in pufB which is ab-
sent from Methylobacterium groups 3 and 4 Rhizobium sp
WL3 and Roseomonas sp WL45 and pufC missing from
Methylobacterium group 4 as well as the novel strain WL4
The remaining genes were found ubiquitously in all strains of
the analysis with only a few exceptions (table 2)
The PGCs were consistently found on the longer contigs of
the initial assemblies of Illumina data produced by SPAdes
(for the strains with fewer than 500 contigs) (table 1) In
Rhizobium sp WL3 and Roseomonas sp WL45 the photosyn-
thetic genes form one continuous cluster with different
however architectures In Alsobacter sp WL4 the genes are
organized in two clusters similar to the Methylobacterium
spp isolates In these cases the genes were found in the
middle of the contigs ruling out the possibility of PGC being
organized into two clusters as an artifact of the assembly
Unique to the architecture present in WL4 is the presence
FIG 2mdashPhylogenomic tree based on the core proteome (CheckM) of the 21 pure bacterial isolates of the analysis including Methylosinus trichosporium
OB3B and Alsobacter sp SH9 The legend shows substitution rates from the root of the tree and the node labels indicated bootstrap support values ()
Zervas et al GBE
2900 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
of bchI before the crtB crtI genes in the first cluster For
Methylobacterium spp groups 2 and 3 show similar gene
syntenymdashthough there are a few differences in gene content
(table 2)mdashwhereas groups 1 and 4 appear to also share the
overall architecture of the PGCs For isolates WL9 and WL19
(both in group 1) the exact location of the crtB and crtI genes
is not known as they appear on a short contig by themselves
Annotating the assemblies on RAST revealed several house-
keeping genes present in the vicinity of these clusters which
further supported the notion of their chromosomal place-
ment The completed genomes resulting from the hybrid as-
semblies (strains WL1 WL3 WL4 and WL45) provided the
final evidence of the exact placement of the PGCs on
the chromosomes of all our isolates The different
architectures of the isolated PGCs are shown in figure 5
General descriptive statistics of the assembled complete
genomes are given in table 3 The hybrid assemblies resulted
in both complete genomes and plasmids Thus Rhizobium sp
WL3 contains two plasmids Roseomonas sp WL45 contains
two chromosomes and seven plasmids Alsobacter sp WL4
with two plasmids Methylobacterium sp WL1 with one
plasmid
Using Roseomonas sp WL45 as outgroup we generated
phylogenetic trees for the 16S acsF bchL bchY crtB pufL
and pufM genes as described previously (data not shown)
We calculated substitution rates from the root of the different
trees including the core proteome (CheckM) tree and com-
pared them to each other (table 4)
Discussion
A New Strategy to Generate Complete Genomes ofUnique Aerobic Anoxygenic Phototrophic Bacteria FromEnvironmental Isolates
Aiming to enhance our knowledge of the prevalence of aer-
obic anoxygenic phototrophic bacteria in the phyllosphere
we designed this study using wheat as the plant of choice
and employed a high-throughput multi-disciplinary approach
to identify isolate and analyze AAP strains From the initial
137 plates we ended up with 4480 bacterial colonies
gt500 of which were probable AAP strains After macroscopic
investigation of size shape color and texture of the colonies
we chose 129 strains for which we ran proteome profiling
using MALDI-TOF MS This approach has been shown to be
FIG 3mdashPairwise proteome comparisons (BLAST-matrix) of the 21 bacterial isolates of the analysis The gray-to-green color gradient indicates low to
high percentages of identity The bottom boxes show the presence of homologs within the same proteome (gray-to-red color gradient shows low to high
presence of homologs in the specific proteome)
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2901
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
able to reproducibly distinguish different taxa and in cases
where reference MALDI-TOF spectra are available in
the database be able to assign Genus andor Species names
(Madsen et al 2015) We then proceeded with whole ge-
nome sequencing on the Illumina Nextseq platform and for
the unique genomes we also proceeded with Oxford
Nanopore Technologies sequencing to generate complete
genomes and plasmids
Following this multi-disciplinary approach that relies on
classical microbiological techniques (plating morphology
etc) biochemical analyses (MALDI-TOF MS) and two differ-
ent types of high-throughput whole genome sequencing we
were able to generate draft and complete genomes of novel
AAP bacteria from the wheat phyllosphere in a relatively quick
(weeks) and efficient manner without losing novel biological
information in the process of reducing redundancy
Bacteriochlorophyll-Containing AAP Bacteria in the WheatPhyllosphere
By investigating the presence of bacteriochlorophylls in the
wheat phyllosphere we were able to identify 21 unique strains
harboring them the majority of which belonged to the genus
Methylobacterium Members of this genus are ubiquitous in
nature and have been detected in both soil and plant surfaces
(Green 2006) Targeting the bchY (chlorophyllide reductase
subunit Y) and pufM (reaction centre subunit M) genes in
phyllosphere samples of different plants it was shown that
the Methylobacteriaceae (alpha-Proteobacteria) comprised
28 of the total pufM sequences from the amplicon sequenc-
ing analysis (Atamna-Ismaeel and Finkel 2012a) Our results
are thus in accordance to these previous studies As
Methylobacteriaceae are indigenous inhabitants of the phyllo-
sphere and have consistently been shown to harbor PGCs it
may be that these photosystems are essential to the
Methylobacteria that inhabit this niche Interestingly in their
study Atamna-Ismaeel et al had negative results from their
bchY amplicon sequencing approach suggesting the absence
of bacteria that contain RC1 (bacteriochlorophyll-a containing
reaction centers) type photosystems Our results show that all
isolates contain the complete cassette of chlorophyllide syn-
thase subunit encoding genes (bchB bchC bchF bchG bchI
bchL bchM bchN bchP bchX bchY bchZ) as well as both the
pufL and pufM genes (table 2) Testing the same primers (Yutin
et al 2009) in silico against our bchY sequences we had pos-
itive results for all of them (data not shown)
Patterns of loss of the carotenoid genes have been ob-
served in other alpha and gamma proteobacteria like in
Rhodobacterales and Sphingomonadales (Zheng et al
2011) The absence of crtA results in the inability to perform
the final step in the spheroidene pathway where hydroxy-
spheroidene is converted to hydroxyspheroidenone
Moreover the absence of crtC in Methylobacterium spp of
groups 3 and 4 completely affects both the spheroidene and
the spirilloxanthin pathways as its product is essential for the
first step of both metabolic pathways starting from neuro-
sporene and lycopene respectively This means that strains
FIG 4mdashWhole genome average nucleotide identity (ANI) cladogram
showing percentages of k-mer identities between the isolates Taxa with
the same color and above 90 identity threshold are considered to belong
in the same species Beijerinckia indica ATCC-9039 was used as outgroup
of the analyses (not defined a priori)
Zervas et al GBE
2902 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Tab
le2
Gen
eC
onte
nt
of
the
Photo
synth
esis
Gen
eC
lust
ers
Iden
tified
inth
e21
Bac
terial
Isola
tes
Bact
eri
och
loro
ph
yll
Syn
thase
Gen
es
Caro
ten
oid
Bio
syn
thesi
sG
en
es
Lig
ht-
Harv
est
ing
Co
mp
lex
Gen
es
Str
ain
acs
Fb
chB
bch
Cb
chF
bch
Gb
chI
bch
Lb
chM
bch
Nb
chP
bch
Xb
chY
bch
Zcr
tAcr
tBcr
tCcr
tDcr
tEcr
tFcr
tIcr
tKp
ucC
pu
fAp
ufB
pu
fCp
ufL
pu
fMp
uh
A
Rh
izo
biu
msp
W
L3
Ro
seo
mo
nas
sp
WL4
5
Als
ob
act
er
sp
WL4
Gro
up
1M
eth
ylo
bact
eri
um
sp
WL9
Meth
ylo
bact
eri
um
sp
WL1
9
Meth
ylo
bact
eri
um
sp
WL6
9
Gro
up
2M
eth
ylo
bact
eri
um
sp
WL1
Meth
ylo
bact
eri
um
sp
WL2
Meth
ylo
bact
eri
um
sp
WL7
Meth
ylo
bact
eri
um
sp
WL1
8
Meth
ylo
bact
eri
um
sp
WL6
4
Gro
up
3M
eth
ylo
bact
eri
um
sp
WL6
Meth
ylo
bact
eri
um
sp
WL3
0
Meth
ylo
bact
eri
um
sp
WL9
3
Meth
ylo
bact
eri
um
sp
WL1
16
Meth
ylo
bact
eri
um
sp
WL1
19
Gro
up
4M
eth
ylo
bact
eri
um
sp
WL8
Meth
ylo
bact
eri
um
sp
WL1
2
Meth
ylo
bact
eri
um
sp
WL1
03
Meth
ylo
bact
eri
um
sp
WL1
20
Meth
ylo
bact
eri
um
sp
WL1
22
NO
TEmdash
Gra
yb
oxe
sin
dic
ate
pre
sen
ceo
fg
en
es
an
dw
hit
eb
oxe
sin
dic
ate
ab
sen
ceo
fg
en
es
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2903
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
of these two groups rely on the zeaxanthin pathway to pro-
duce carotenoid pigments and most likely nostoxanthin and
erythroxanthin which however do not participate in the
light-harvesting process but rather have a photoprotection
role (Noguchi et al 1992 Yurkov and Csotonyi 2009)
Phylogenetic Observations
In our study we sequenced and analyzed 21 aerobic anoxy-
genic photosynthetic bacteria most of which belonged to the
genus Methylobacterium the presence of which in the
environment has already been discussed previously
Methylobacterium spp are facultative methylotrophs and
can utilize a variety of C1 C2 C3 and C4 compounds
Such compounds are often found in the phyllosphere due
to plant metabolism (Vorholt 2012 Remus-Emsermann
et al 2014) Methylobacterium spp show a wide range in
chromosome size and number of plasmids they contain
(Marx et al 2012) The genus is represented well in
GenBank with 100 complete or draft genomes (last accessed
April 2019) However most of the sequences do not have
FIG 5mdashArchitecture of the different photosynthetic gene clusters Arrows indicate the orientation of the genes Sizes are not representatives of gene
length Only the genes shared by all PGCs were shown Genes that may be missing from some strains are denoted with parentheses on the figurersquos legend
Table 3
General Descriptive Statistics of the Four Complete AAP Bacterial Genomes
Taxon Genome Size (bp) GC Number of Genes Plasmids Plasmids Size (bp) GC
Rhizobium sp WL3 4568855 6160 4450 1 700631 586
mdash mdash mdash 2 85350 575
Roseomonas sp WL45 4533887 708 5042 1 262815 655
1011854 707 1138 2 240556 662
mdash mdash mdash 3 152922 658
mdash mdash mdash 4 85279 646
mdash mdash mdash 5 84774 661
mdash mdash mdash 6 83273 656
mdash mdash mdash 7 65860 648
Alsobacter sp WL4 5405668 679 5013 1 27178 620
mdash mdash mdash 2 9451 606
Methylobacterium sp WL1 6215463 691 6431 1 35731 638
Zervas et al GBE
2904 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
species affiliation (fig 4) Thus we were not able to assign any
of the 18 isolates to specific Methylobacterium species apart
perhaps from group 2 (WL1 WL2 WL7 WL18 WL64) which
is placed as sister to M brachiatum in the MASH-ANI analysis
Even in this case though M pseudosasicola BL36 is also
placed in the same clade as M brachiatum which raises the
question of proper nomenclature This is further evidenced in
the same tree (cyanide clade below) where seven species
share gt90 of their genome For the remaining isolates
WL19 is closely affiliated to other Methylobacterium strains
that have not been identified A similar picture is shown for
WL69 On the other hand WL19 appears to be a novel spe-
cies with unique genomic features and so do the strains from
groups 3 and 4 (red clade) Thus despite how ubiquitous
Methylobacterium is in nature it is clear that part of its diver-
gence is still missing Moreover a critical evaluation of the
current nomenclature of the genus would be beneficial based
on the whole genome sequences because closely related
genomes (MASH-ANI identitygt 97) have been given differ-
ent species affiliations
Apart from this genus we also identified a Rhizobium sp
a Roseomonas sp and a novel Alsobacter strain (WL4) that
belongs in the Methylocystaceae a family of methanotroph
bacteria This is the first time that AAP bacteria from this
family have been isolated and their complete genomes assem-
bled Its closely related Alsobacter spp have been shown to
exhibit high Thallium tolerance (Bao et al 2006) and it will be
interesting to identify heavy-metal resistance genes in non-
heavy-metal polluted phyllosphere samples It is worth men-
tioning that if WL4 contains heavy-metal resistance genes on
top of the PGC this will have been shown for the first time
and its potential to bioremediate heavy-metal contaminated
areas if applied as plant growth promotion and environmental
remediation agent is of high value and needs to further be
explored
Molecular Evolution of the PGCs
For all gene trees of the PGC we observed that Rhizobium sp
WL3 evolves faster than the other isolates followed by
Table 4
Substitutions Per Site in the Different Phylogenetic Analyses
Strain 16S acsF bchL bchY crtB pufL pufMsuper-matrix
CheckM
Roseomonas sp WL45 0096 0147 0100 0235 0250 0095 0132 0160 0184
Rhizobium sp WL3 0184 0622 0767 0690 1370 0343 0525 0580 0372
Alsobacter sp WL4 0151 0462 0233 0490 0870 0258 0334 0380 0403
Methylobacterium sp WL9 0226 0418 0372 0420 0670 0270 0319 0364 0477
Methylobacterium sp WL19 0221 0413 0300 0460 0600 0233 0346 0364 0468
Methylobacterium sp WL69 0231 0373 0333 0470 0620 0260 0358 0368 0468
Methylobacterium sp WL1 0233 0422 0344 0485 0480 0280 0385 0380 0459
Methylobacterium sp WL2 0233 0422 0344 0485 0480 0280 0385 0380 0459
Methylobacterium sp WL7 0238 0427 0344 0480 0480 0278 0385 0380 0459
Methylobacterium sp WL18 0238 0427 0344 0480 0480 0280 0385 0380 0459
Methylobacterium sp WL64 0233 0422 0344 0480 0485 0280 0385 0380 0459
Methylobacterium sp WL6 0199 0378 0256 0425 0680 0208 0311 0340 0455
Methylobacterium sp WL30 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL93 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL116 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL119 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL8 0188 0373 0258 0430 0680 0230 0358 0348 0459
Methylobacterium sp WL12 0188 0373 0256 0430 0680 0230 0358 0348 0459
Methylobacterium sp WL103 0188 0373 0256 0430 0850 0230 0358 0364 0459
Methylobacterium sp WL120 0188 0373 0258 0430 0680 0233 0358 0348 0459
Methylobacterium sp WL122 0188 0373 0256 0430 0680 0230 0358 0348 0459
Susbstitutions per site
Gro
up1
Gro
up 2
Gro
up 3
Gro
up 4
NOTEmdashAll values correspond to distances from the root where Roseomonas sp WL45 was chosen as outgroup for all the alignments
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2905
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Alsobacter sp WL4 Both these isolates also show much
higher substitution rates compared with their respective 16S
genes and core proteomes (checkM) which indicates that the
PGCs are under relaxed selection pressure In all instances the
phylogenetic trees constructed showed the same topology
and the same grouping of the Methylobacterium isolates
which suggests that the PGCs were incorporated in the
genomes of these strains relatively early in their evolution If
the PGC was free to move laterally then we would have ob-
served very different phylogenetic trees for its genes com-
pared with the 16S and core proteome analyses
Interestingly these observations do not reflect on the archi-
tecture of the PGC where Roseomonas sp WL45 and
Rhizobium sp WL3 contain the PGC in one continuous re-
gion whereas Alsobacter sp WL4 and Methylobacterium sp
WL1 have it in two pieces
For the 18 Methylobacterium isolates group 1 and 2
strains (WL9 WL19 WL69 and WL1 WL2 WL7 WL18
WL64) evolve consistently at different rates compared
with the groups 3 and 4 It appears that there is relatively
strong selection pressure on bchL pufL and pufM which
in all cases are more similar to each other and also show
slower molecular evolution rates compared with ascF
bchY and crtB which in all Methylobacterium isolates
evolve at least twice as fast compared with the 16S
gene (table 4) They also show fewer substitutions per
site compared with the core proteome analysis This result
partially contrasted with previously adopted methods of
using bchY as a marker gene for RC1 clusters (eg
Atamna-Ismaeel and Finkel 2012a) On the other hand
the use of pufM appears more sensible It would how-
ever be more suitable to investigate the possibility of us-
ing other genes for metagenome amplicon sequencing
approaches such as pufL or bchL which show a smaller
degree of divergence in different genera thus universal
primers might be more successful in detecting a wide va-
riety of AAP bacteria in environmental samples
Interestingly while Groups 1 and 2 isolates exhibit higher
substitution rates of the PGC genes compared with
Groups 3 and 4 this is not the case for crtB which evolves
a lot slower and also pufM which shows similar molec-
ular evolution in the genus Overall it seems that the
PGCs were incorporated in ancestral strains that later di-
verged to give rise to the different AAP taxa This is fur-
ther corroborated by the fact that almost all phylogenetic
trees (eg fig 5) are consistent and in agreement with
both the 16S phylogenetic tree (fig 1) as well as the phy-
logenomic tree (fig 2) If the genes of the PGCs had been
transferred to these AAP several times there would have
been significant differences both in the sequence level
and the phylogenetic analysis of the different genes
which would be expected to result in different topologies
and diverse distances from the root even for closelymdash
based on 16S analysismdashrelated strains
The Significance of AAP Bacteria in Wheat Phyllosphere
In this study we isolated for the first time a variety of AAP
bacteria that are present in the wheat phyllosphere adding to
the metagenomics knowledge provided earlier from five other
plants (Atamna-Ismaeel and Finkel 2012a 2012b) The phyllo-
sphere is a rather harsh environment with little water avail-
ability scarce resources very few suitable locations extreme
temperature difference between day and night and high
dosages of UV radiation (Vorholt 2012) AAP bacteria have
the ability to utilize light in photophosphorylation processes
through which they produce ATP and at the same time funnel
the few available nutrients in other metabolic pathways
(Yurkov and Hughes 2017) This has been shown to give
them an edge in vitro after exposure to light compared with
nonAAP bacteria (Ferrera et al 2011) Considering that phyl-
losphere bacteria protect the plant from pathogenic microbes
by secreting metabolites and by occupying the available
niches being able to persevere and outgrow other bacteria
gives a significant edge to AAP taxa and probably points
toward the suggestion of positive selection of such bacteria
in their phyllosphere by the plants themselves as they appear
to be absent from the soil (Atamna-Ismaeel and Finkel
2012b) At this point it is not clear exactly how AAP bacteria
benefit the plant especially since the phyllosphere is a com-
plex habitat with an ever-changing harsh environment In the
past years great efforts have been carried out on how to
improve crop production by studying the rhizosphere It has
also been suggested that the phyllosphere communities play a
significant role in plant health and growth which ultimately
leads to increased yields (Lindow and Brandl 2003) Given the
current projections of world population growth and the sup-
ply of food (UN 2017) it becomes evident that more in-depth
studies investigating the phyllosphere of plants and especially
crops and identifying the underlying drivers of the bacterial
communities and their interactions with their plant hosts is of
paramount importance Thus by isolating maintaining and
analyzing the genomes of phyllosphere isolates we may iden-
tify strains that may prove useful as plant growth promoting
agents in future field trials investigating the effect of phyllo-
sphere inoculation on plant growth health and yield
Supplementary Material
Supplementary data are available at Genome Biology and
Evolution online
Acknowledgments
The authors would like to thank Tue Sparholt Jorgensen for
collecting the wheat sample The authors also want to thank
Tina Thane (AU) Tanja Begovic (AU) and Margit Frederiksen
(NRCWE) for capable laboratory assistance We thank Michal
Koblızek for providing the initial model of the colony infrared
imaging system This work was funded by a Villum
Zervas et al GBE
2906 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Experiment grant (no 17601) and a Marie Skłodowska-Curie
AIAS-COFUND fellowship (EU-FP7 program Grant
Agreement No 609033) awarded to YZ
Author Contributions
AZ YZ LHH designed the study YZ isolated the strains
performed colony infrared imaging and extracted DNA for
Illumina sequencing YZ and AMM performed the MALDI-
TOF MS analysis AZ extracted DNA and performed ONT
Sequencing AZ carried out the bioinformatics and phyloge-
netic analyses AZ and YZ drafted the manuscript AZ
YZ AMM and LHH revised the manuscript for
intellectual content All authors read and approved the
manuscript
Literature Cited
Atamna-Ismaeel N Finkel O 2012a Bacterial anoxygenic photosynthesis
on plant leaf surfaces Environ Microbiol Rep 4(2)209ndash216
Atamna-Ismaeel N Finkel OM 2012b Microbial rhodopsins on leaf sur-
faces of terrestrial plants Environ Microbiol 14(1)140ndash146
Aziz RK et al 2008 The RAST Server rapid annotations using subsystems
technology BMC Genomics 9(1)75
Bankevich A et al 2012 SPAdes a new genome assembly algorithm and
its applications to single-cell sequencing J Comput Biol
19(5)455ndash477
Bao Z Sato Y Kubota M Ohta H 2006 Isolation and characterization of
thallium-tolerant bacteria from heavy metal-polluted river sediment
and non-polluted soils Microb Environ 21(4)251ndash260
Beatie GA 2002 Leaf surface waxes and the process of leaf colonization
by microorganisms In Lindow SE Hecht-Poinar EI Elliott VJ editors
Phyllosphere Microbiology St Paul (USA) APS Press pp 3ndash26
Beattie GA Lindow SE 1999 Bacterial colonization of leaves a spectrum
of strategies Phytopathology 89(5)353ndash359
Brettin T et al 2015 RASTtk a modular and extensible implementation of
the RAST algorithm for building custom annotation pipelines and an-
notating batches of genomes Sci Rep 58365
Carvalho SD Castillo JA 2018 Influence of light on plantndashphyllosphere
interaction Front Plant Sci 91482
Ferrera I Gasol JM Sebastian M Hojerova E Koblızek M 2011
Comparison of growth rates of aerobic anoxygenic phototrophic bac-
teria and other bacterioplankton groups in coastal Mediterranean wa-
ters Appl Environ Microbiol 77(21)7451ndash7458
Ferrera I Sanchez O Kolarova E Koblızek M Gasol JM 2017 Light
enhances the growth rates of natural populations of aerobic anoxy-
genic phototrophic bacteria ISME J 11(10)2391ndash2393
Gorlach K Shingaki R Morisaki H Hattori T 1994 Construction of eco-
collection of paddy field soil bacteria for population analysis J Gen
Appl Microbiol 40(6)509ndash517
Gourion B Rossignol M Vorholt JA Lindow SE 2006 A proteomic study
of Methylobacterium extorquens reveals a response regulator essential
for epiphytic growth wwwintegratedgenomicscom Accessed
November 26 2018
Green PN 2006 Methylobacterium In Dworkin M Falkow S Rosenberg
E Schleifer K Stackebrandt E editors The prokaryotes ndash a handbook
on the biology of bacteria New York Springer p 257ndash265
Hauruseu D Koblızek M 2012 Influence of light on carbon utilization in
aerobic anoxygenic phototrophs Appl Environ Microbiol
78(20)7414ndash7419
Katoh K Misawa K Kuma K Miyata T 2002 MAFFT a novel method for
rapid multiple sequence alignment based on fast Fourier transform
Nucleic Acids Res 30(14)3059ndash3066
Knief C et al 2012 Metaproteogenomic analysis of microbial communi-
ties in the phyllosphere and rhizosphere of rice ISME J
6(7)1378ndash1390
Knief C Frances L Vorholt JA 2010 Competitiveness of diverse methyl-
obacterium strains in the phyllosphere of arabidopsis thaliana and
identification of representative models including M extorquens
PA1 Microb Ecol 60(2)440ndash452
Kwak MJ et al 2014 Genome information of Methylobacterium oryzae a
plantndashprobiotic methylotroph in the phyllosphere PLoS One
9(9)e106704
Lindow SE Brandl MT 2003 Microbiology of the phyllosphere Appl
Environ Microbiol 69(4)1875ndash1883
Madsen AM Zervas A Tendal K Nielsen JL 2015 Microbial diversity in
bioaerosol samples causing ODTS compared to reference bioaerosol
samples as measured using Illumina sequencing and MALDI-TOF
Environ Res 140255ndash267
Martin M 2011 Cutadapt removes adapter sequences from high-
throughput sequencing reads EMBnet J 17(1)10ndash12
Marx CJ et al 2012 Complete genome sequences of six strains of the
genus Methylobacterium b J Bacteriol 194(17)4746
Noguchi T Hayashi H Shimada K Takaichi S Tasumi M 1992 In vivo
states and functions of carotenoids in an aerobic photosynthetic bac-
terium Erythrobacter longus Photosynth Res 31(1)21ndash30
Olm MR Brown CT Brooks B Banfield JF 2017 dRep a tool for fast and
accurate genomic comparisons that enables improved genome recov-
ery from metagenomes through de-replication ISME J
11(12)2864ndash2868
Overbeek R et al 2014 The SEED and the Rapid Annotation of microbial
genomes using Subsystems Technology (RAST) Nucl Acids Res
42(D1)D206ndashD214
Parks DH Imelfort M Skennerton CT Hugenholtz P Tyson GW 2015
CheckM assessing the quality of microbial genomes recovered from
isolates single cells and metagenomes Genome Res 25(7)1043ndash1055
Remus-Emsermann MNP et al 2014 Spatial distribution analyses of nat-
ural phyllosphere-colonizing bacteria on Arabidopsis thaliana revealed
by fluorescence in situ hybridization Environ Microbiol
16(7)2329ndash2340
Stamatakis A 2006 RAxML-VI-HPC maximum likelihood-based phyloge-
netic analyses with thousands of taxa and mixed models
Bioinformatics 22(21)2688ndash2690
UN 2017 World population prospects the 2017 revision Data Booklet
httpspopulationunorgwppPublicationsFilesWPP2019_DataBook-
letpdf
Vesth T Lagesen K Acar euroO Ussery D 2013 CMG-biotools a free work-
bench for basic comparative microbial genomics PLoS One
8(4)e60120
Vorholt JA 2012 Microbial life in the phyllosphere Nat Rev Microbiol
10(12)828ndash840
Wick RR Judd LM Gorrie CL Holt KE 2017 Unicycler resolving bacterial
genome assemblies from short and long sequencing reads PLoS
Comput Biol 13(6)e1005595
Wilson M Lindow SE 1994a Coexistence among epiphytic bacterial pop-
ulations mediated through nutritional resource partitioning Appl
Environ Microbiol 604468ndash77
Wilson M Lindow SE 1994b Ecological similarity and coexistence of epi-
phytic ice-nucleating (ice) pseudomonas syringae strains and a non-
ice-nucleating (ice) biological control agent Appl Environ Microbiol
60(9)3128ndash3137
Woodward FI Lomas MR 2004 Vegetation dynamics ndash simulating
responses to climatic change Biol Rev 79(3)643ndash670
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2907
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Xu J Gordon JI 2003 Honor thy symbionts Proc Natl Acad Sci USA
100(18)10452ndash10459
Yurkov V Csotonyi JT 2009 New light on aerobic anoxygenic photo-
trophs In Hunter NC Daldal F Thurnauer MC Beatty JT editors
The purple photosynthetic bacteria Dordrecht Springer thorn Business
Media BV p 31ndash55 doi101007978-1-4020-8815-5_3
Yurkov V Hughes E 2017 Aerobic Anoxygenic Phototrophs four
decades of mystery In Hallenbeck PC editor Modern topics in the
phototrophic prokaryotes environmental and applied aspects p
193ndash214
Yutin N et al 2009 BchY-based degenerate primers target all types of
anoxygenic photosynthetic bacteria in a single PCR Appl Environ
Microbiol 75(23)7556ndash7559
Zeng Y Feng F Medova H Dean J Koblizek M 2014 Functional
type 2 photosynthetic reaction centers found in the rare bacterial phy-
lum Gemmatimonadetes Proc Natl Acad Sci USA 111(21)7795ndash7800
Zheng Q et al 2011 Diverse arrangement of photosynthetic gene
clusters in aerobic anoxygenic phototrophic bacteria PLoS One
6(9)e25050
Associate editor Esperanza Martinez-Romero
Zervas et al GBE
2908 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
- evz204-TF1
- evz204-TF2
-
Genomics of Aerobic Photoheterotrophs in Wheat
Phyllosphere Reveals Divergent Evolutionary Patterns of
Photosynthetic Genes in Methylobacterium spp
Athanasios Zervas 1 Yonghui Zeng12 Anne Mette Madsen3 and Lars H Hansen141Section of Environmental Microbiology and Biotechnology Department of Environmental Science Aarhus University Roskilde Denmark2Aarhus Institute of Advanced Studies Aarhus University Denmark3The National Research Centre for the Working Environment Copenhagen Denmark4Environmental Microbial Genomics Group Department of Plant and Environmental Sciences University of Copenhagen Frederiksberg Denmark
Corresponding authors E-mails zengenvsaudk lhhaplenkudk
Accepted September 11 2019
Data deposition This project has been deposited at GenBank under the accession numbers SAMN12513103ndashSAMN12513123
Abstract
Phyllosphere is a habitat to a variety of viruses bacteria fungi and other microorganisms which play a fundamental role in
maintaining the health of plants and mediating the interaction between plants and ambient environments A recent addition to
this catalogueofmicrobialdiversitywas theaerobicanoxygenicphototrophs (AAPs) agroupofwidespreadbacteria thatabsorb light
through bacteriochlorophyll a (BChl a) to produce energy without fixing carbon or producing molecular oxygen However culture
representatives of AAPs from phyllosphere and their genome information are lacking limiting our capability to assess their potential
ecological roles in this unique niche In this study we investigated the presence of AAPs in the phyllosphere of a winter wheat
(Triticum aestivum L) in Denmark by employing bacterial colony based infrared imaging and MALDI-TOF mass spectrometry (MS)
techniques A total of 4480 colonies were screened for the presence of cellular BChl a resulting in 129 AAP isolates that were
further clustered into 21 groups based on MALDI-TOF MS profiling representatives of which were sequenced using the Illumina
NextSeq and Oxford Nanopore MinION platforms Seventeen draft and four complete genomes of AAPs were assembled belonging
in Methylobacterium Rhizobium Roseomonas and a novel Alsobacter We observed a diverging pattern in the evolutionary rates of
photosynthesis genes among the highly homogenous AAP strains of Methylobacterium (Alphaproteobacteria) highlighting an
ongoing genomic innovation at the gene cluster level
Key words AAP bacteria bacteriochlorophyll phyllosphere MALDI-TOF MS Nanopore photosystems
Introduction
Plantndashmicrobe interactions both above (phyllosphere) and
below (rhizosphere) ground are common in nature
Traditionally these relationships are investigated in the rhizo-
sphere where conditions are relatively stable and nutrient
availability is rather high (Vorholt 2012 Carvalho and
Castillo 2018) Despite recent work (reviewed in eg
Lindow and Brandl 2003 Vorholt 2012) the microbendash
phyllosphere (which is comprised by the aerial partsmdashand
especially the leavesmdashof plants) interactions remain relatively
understudied compared with rhizosphere Models suggest
that the total leaf surface is gt1 billion km2 (Woodward and
Lomas 2004) potentially colonized by up to 1026 bacterial
cells (Lindow and Brandl 2003)
Although the phyllosphere seems to be a common envi-
ronment for bacteria to thrive in it may be ldquoextremerdquo for a
plethora of reasons First different plants exhibit different
growth patterns and climate adaptations Annual plants com-
plete their life cycle in just one growth season whereas pe-
rennial plants grow and shed leaves every year This creates a
discontinuous ever-changing habitat (Vorholt 2012)
Meanwhile environmental changes also affect the phyllo-
sphere and its inhabitants Winds rainfall frost and drought
all play important roles in shaping the conditions encountered
The Author(s) 2019 Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (httpcreativecommonsorglicensesby40) which permits unrestricted reuse
distribution and reproduction in any medium provided the original work is properly cited
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2895
GBED
ownloaded from
httpsacademicoupcom
gbearticle-abstract111028955570988 by Royal Library C
openhagen University user on 16 January 2020
by microorganisms of the phyllosphere However the most
significant environmental driver is sunlight Temperature dif-
ferences in the upper leaves especially those that form the
canopy of various habitats may range up to 50 C between
day and night Sunlight though beneficial for photosynthetic
organisms is also comprised of UV-radiation which is dam-
aging to the DNA of both prokaryotes and eukaryotes
(Remus-Emsermann et al 2014) In addition to the abiotic
factors that make the phyllosphere a much more hostile en-
vironment than the rhizosphere its inhabitants also facemdash
directly or indirectlymdashthe influence of their plant host
Leaves are surrounded by a thin laminar and waxy layer
the cuticle which renders the leaf surface hydrophobic
thus removing the excess of water that is collected due to
rainfall dew or respiration from the stomata This leads to the
retention of little water most usually in veins and other cavities
of the cuticle These microformations also protect the colo-
nizers from the surrounding environment (Beatie 2002) Apart
from niche competition and scarce water availability phyllo-
sphere colonizers also need to compete for nutrients as the
cuticle makes the surface virtually impermeable to nutrients
deriving from diffusion through the cells of the plant host
(Wilson and Lindow 1994a 1994b) At the same time they
also need to protect themselves from potential invaders and a
plethora of antimicrobial compounds of prokaryotic or eu-
karyotic origin (Vorholt 2012)
Albeit the harsh conditions they encounter phyllo-
sphere microogransims exhibit remarkable biodiversity
which in certain cases can be comparable to that of hu-
man gut microbiome (Xu and Gordon 2003 Knief et al
2012) Through recent genomics and metagenomics stud-
ies several bacterial genera such as Methylobacterium
have been shown to be ubiquitous in the phyllosphere
(Knief et al 2010) and their role in nutrient recycling plant
growth promotion and protection has been evidenced pre-
viously (Beattie and Lindow 1999 Gourion et al 2006
Atamna-Ismaeel and Finkel 2012a Kwak et al 2014)
One of the most interesting recent findings in phyllosphere
microbiota was the presence of aerobic anoxygenic photo-
trophic (AAP) bacteria and rhodopsin-harboring bacteria in
a variety of land plants (Atamna-Ismaeel and Finkel 2012a
2012b) AAP bacteria are commonly found in aquatic envi-
ronments ranging from the arctic to the tropics and from
high salinity lakes to pristine high altitude lakes (Yurkov and
Hughes 2017) They rely on organic carbon compounds to
cover their nutritional requirements and can also utilize
light to produce energy in the form of ATP without fixing
carbon or producing oxygen AAPs have been found in a
variety of extreme environments and have shown their po-
tential to outgrow nonAAP bacteria under nutrient limiting
conditions (Hauruseu and Koblızek 2012 Ferrera et al
2017) Thus their presence in the phyllosphere may not
be surprising However their relative abundance at least
in the rice phyllosphere (Oryza sativa) was up to three
times higher than what is commonly found in marine eco-
systems (Atamna-Ismaeel and Finkel 2012a)
These early culture-independent metagenomics studies
provided valuable information about the presence of AAP
bacteria in the phyllosphere However no pure cultures of
AAPs have been described so far from phyllosphere and
thus detailed genome information is lacking which prevents
an in-depth understanding of the ecological roles and evolu-
tionary trajectory of AAPs in this unique niche To expand our
genomics views of this ecologically important group of bac-
teria in the phyllosphere we designed this study with the
following aims 1) to create the first collection of AAP isolates
from the phyllosphere 2) to characterize their photosynthesis
gene clusters (PGCs) and compare their gene contents and
molecular evolutionary patterns and 3) to expand the data-
base of complete genomes of AAP bacteria from nonaquatic
environments By combining colony infrared (IR) imaging
MALDI-TOF mass spectrometry (MS) and Illumina and
Oxford Nanopore sequencing technologies we were able to
provide 20 draft genomes that contain complete PGCs and
four complete genomes that contain plasmids The further
comparison of the evolutionary trajectories of their PGCs
revealed a diverging pattern among the highly homogenous
Methylobacterium strains highlighting an ongoing genomic
innovation at the gene cluster level
Materials and Methods
Sample Collection
A whole intact plant of winter wheat (Triticum aestivum L)
with a height of circa 60 cm was collected from a field in
Roskilde Denmark on June 6 2018 and transported to the
lab for bacterial isolation on the same day Wheat leaves were
cut off and cleaned with running sterile water to remove dust
and other temporary deposits from the ambient environment
Microbiota were collected from eight pieces of leaves by rins-
ing the leaves in 80 ml PBS solution (pH 74) in a Oslash 150-mm
petri dish and scraping the leave surface repeatedly with ster-
ile swabs until the PBS solution slightly turned greenish due to
the fall-off of leave cells From the supernatant 14 ml were
plated on 15 strength R2A plates (resulting in a total of 137
agar plates) and incubated at room temperature with normal
indoor light conditions for two weeks AAP bacteria were
identified using a custom screening chamber equipped with
an IR imaging system described in detail previously (Zeng
et al 2014) In short plates were placed in a chamber sealed
from light and exposed to a source of green lights AAP bac-
teria containing bacteriochlorophyll absorb the green light
and emit radiation in the near IR spectrum which was cap-
tured by a NIR sensitive CMOS camera The image was proc-
essed and the glowing colonies indicated the presence of
BChl a The colonies that give positive signals were marked
restreaked onto 15 strength R2A plates and incubated at
room temperature for 72 h Verification of the light absorbing
Zervas et al GBE
2896 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
capabilities of the isolated strains was performed using the
same setup Strains that passed the verification step were
further restreaked onto 15 strength R2A plates to ensure
pure colonies of single AAP strains
Selection of AAP Strains and Whole Genome Sequencing
The isolated AAP strains were grouped and dereplicated using
a MALDI-TOF MS (Microflex LT Bruker Daltonics Bremen
Germany) before performing whole genome sequencing to
reduce the cost while maintaining the biodiversity Briefly a
toothpick was used to transfer a small amount of an AAP
bacterial colony onto the target plate (MSP 96 polished steel
Bruker) that was evenly spread out and formed a thin layer of
biomass on the steel plate The sample was then overlaid with
70 formic acid and allowed for air dry before addition of
1lL MALDI-MS matrix solution (a-cyano-4-hydroxycinnamic
acid SigmandashAldrich) A bacterial standard with well-
characterized peaks (Bruker Daltonics) was used to calibrate
the instrument The standard method ldquoMBT_AutoXrdquo was ap-
plied to obtain proteome profiles within the mass range of 2ndash
20 kDa using the flexControl software (Bruker) The
flexAnalysis software (Bruker) was used to smooth the data
subtract baseline and generate main spectra (MSP) followed
by a hierarchical clustering analysis with the MALDI Biotyper
Compass Explorer software which produced a dendrogram
as the output for visual inspection of similarities between
samples An empirical distance value of 50 was used as the
cutoff for defining different groups on the dendrogram cor-
responding to different strainsspecies
From each group on the dendrogram one isolate was cho-
sen and restreaked on 1=2 R2A plates and incubated at room
temperature for one week Prior to DNA isolation the plates
were again tested for the presence of light absorbing pigments
as previously described Bacterial colonies were scraped from
the plates using a loop and immersed in 400mL PBS solution
vortexed until the cells had separated spun down in a table-top
centrifuge at 10000 g at 4 C the supernatant was removed
and the pellets were redissolved in milliQ H2O High molecular
weight DNA was extracted from each strain using a modified
Masterpure Complete DNA amp RNA Purification Kit (Lucigen) by
replacing the elution buffer with a 10mM TrisndashHClmdash50mM
NaCl (pH 75ndash80) solution at the final step DNA concentration
was quantified on a Qubit 20 (Life Technologies) using the
Broad Range DNA Assay kit and DNA quality was measured
on a Nanodrop 2000C (Thermo Scientific) Whole genome
shotgun sequencing was performed on all AAP strains on the
Illumina Nextseq platform in house using the 2150bp chem-
istry Selected strains were also sequenced on the MinION plat-
form (Oxford Nanopore Technologies UK) in house using the
Rapid Barcoding Kit (SBQ-RBK004) and the FLO-MIN106 flow
cell (R941) following the manufacturerrsquos instructional
manuals
Genome Assembly and Analyses
The Illumina pair end reads were trimmed for quality and
ambiguities using Cutadapt (Martin 2011) and assembled
using SPAdes (Bankevich et al 2012) The resulting assem-
blies were analyzed using dRep (Olm et al 2017) to compare
their average nucleotide identities (ANI) in order to identify
clusters of closely related taxa The assemblies were
imported in Geneious R1125 (Biomatters Ltd) and PGCs
were annotated using the Roseobacter litoralis strain
Och149 plasmid pRL0149 (CP002624) as reference under
relaxed similarity criteria (40) The suggested genes of
probable PGCs were analyzed using BLAST (megablast
BlastN) to verify their annotations For full genome annota-
tions the assemblies were uploaded on RAST (Aziz et al
2008 Overbeek et al 2014 Brettin et al 2015) The pre-
dicted translated protein coding genes were imported in
CMG-Biotools (Vesth et al 2013) for pairwise proteome
comparisons using the native blastmatrix program
Hybrid assemblies of pair-end Illumina reads and long
Oxford Nanopore Technologies reads were performed using
the Unicycler assembler (Wick et al 2017) utilizing the bold
assembly mode The resulting complete genomes were
imported in Geneious where their circularity was confirmed
by mapping-to-reference runs using the short and long reads
from the respective hybrid assemblies and the native
Geneious mapper under default settings in the Mediumndash
Low sensitivity mode
Phylogenetic Analyses
Identification of the 16S ribosomal DNA sequences was per-
formed by rnammer in CMG-Biotools Data sets containing
the different genes from the identified PGCs were created in
Geneious The resulting sequences were aligned using MAFFT
v7388 (Katoh et al 2002) and it FFT-NS-I x1000 algorithm
with default settings as implemented in Geneious
Phylogenetic trees were created using RAxML (Stamatakis
2006) employing the GTR-GAMMA nucleotide substitution
model and the rapid bootstrapping and search for best scor-
ing ML-tree with 100 bootstraps replicates and starting from
a random tree Analysis of the PGCs was performed on five
selected genes that were present in all isolates namely acsF
bchL bchY crtB pufL and pufM Individual gene alignments
and phylogenetic trees were performed as described above
A super-matrix (8240 characters) comprised of the total
seven gene alignments was created using the built-in
ldquoConcatenate alignmentsrdquo function in Geneious and a phy-
logenetic tree was created as above All trees were visually
inspected for disagreements Finally a phylogenomic tree
was constructed in RAxML employing the GAMMA
BLOSUM62 substitution model and the rapid bootstrapping
and search for best scoring ML-tree algorithm with 100 boot-
straps replicates and starting from a random tree using as
input the core protein alignment consisting of 6988
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2897
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
characters created with CheckM (Parks et al 2015) and its
lineage_wf algorithm with default settings Substitution rates
were calculated as distances from the root using the legend
provided by the phylogenetic trees
Results
Identification and Isolation of AAP Bacteria from thePhyllosphere
The 129 phyllosphere isolates that were tested positive for the
presence of bacteriochlorophyll-a were placed in 21 groups
according to their protein mass profiles based on MALDI-TOF
MS (supplementary fig MALDI-TOF Supplementary Material
online) A representative from each group was sequenced
(Illumina) and their genomes assembled (SPAdes) WL1 and
WL2 were from the same MALDI-TOF group and their
genomes are identical (ANI 100) confirming the validity
of MALDI-TOF MS for grouping similar isolates Information
about their phylogeny total genome size and genome as-
sembly statistics are presented in table 1
Analysis of the 16S genes showed that 17 isolates
belonged in Methylobacterium (separated in four distinct
groups) whereas the remaining three belonged to
Rhizobium Roseomonas and one novel species was affiliated
to the Methylocystaceae family (WL4) with most hits belong-
ing to uncultured environmental strains The only character-
ized hits of WL4 belonged to Methylosinus trichosporium
(Gorlach et al 1994) and Alsobacter metallidurans a recently
characterized novel genus novel species (Bao et al 2006)
However in both instances the similarity was low
(985) We included strains of both species and created
a 16S phylogenetic tree (fig 1) which placed Alsobacter sp
WL4 as sister to the Alsobacter-Methylosinus clade
The core proteome analysis (fig 2mdashCheckM) including M
trichosporium OB3b and Alsobacter sp SH9 established the
phylogeny of strain WL4 and it was highly supported as sister
to Alsobacter sp SH9 (BSfrac14 100) both of which were
distinct from M trichosporium OB3b (BSfrac14 100)
Methylobacterium strains were still placed into four distinct
groups (Group 1 WL9 WL19 WL69 Group 2 WL1 WL2
WL7 WL18 WL64 Group 3 WL6 WL30 WL93 WL116
WL119 Group 4 WL8 WL12 WL103 WL120) This was
further supported by the whole proteome comparison
(fig 3mdashBLAST matrix) For Groups 3 and 4 the average
core protein content is 4500 proteins whereas for Group
2 it is close to 4000 This may signify the presence of
either more or larger plasmids that characterize the strains
of this group For Group 1 this number falls even more as
in this case there clearly are three distinct species that make
up the group
We expanded the analysis to include 100 complete and
draft Methylobacterium spp genomes available in GenBank
(last assessed April 2019) and also other representatives from
the Methylocystaceae aiming at further establishing strain
WL4 in the genus Alsobacter and also providing greater res-
olution on the Methylobacterium isolates with a special focus
on the divergent group 1 Information about the included
genomes is given in supplementary table 1 Supplementary
Material online We performed whole genome comparisons
using ANI (fig 4mdashdRep)
The analysis showed that Group 1 isolates belonged in
different Methylobacterium species groups 3 and 4 were dif-
ferent strains of the same species (MASH ANI identity be-
tween groupsgt 90) whereas group 3 isolates are closely
related strains possibly belonging to Methylobacterium bra-
chiatum Strain WL4 is again placed close to the two identical
Alsobacter genomes (Alsobacter sp SH9 and A metallidurans
PVZS01) far away from the three remaining species from
within the Methylocystaceae (M trichosporium OB3b
Methylocystis rosea GW6 Methylocystis bryophila S285)
However the MASH-ANI identity between WL4 and
Alsobacter is below the 80 threshold of genus assignment
Therefore we took the top 200 megablast hits for WL4rsquos 16S
region (information provided in supplementary table 2
Supplementary Material online) aligned them including strain
WL4 as mentioned previously and created a phylogenetic tree
of the 201 sequences (supplementary fig 1 Supplementary
Material online) The results clearly show that WL4 belongs in
Alsobacter whereas Methylosinus spp form their own dis-
tinct cluster
Table 1
Genome Statistics and Phylogenetic Information of the 21 Isolates that
were Illumina Sequenced in This Study
Strain Taxonomy Contigs Genome Size N50 GC
WL1 Methylobacterium 654 6258106 19384 6892
WL2 Methylobacterium 298 6292310 52057 6899
WL3 Rhizobium 55 5333961 210875 6114
WL4 Alsobacter 139 5412272 86934 6792
WL6 Methylobacterium 708 5549874 15034 6972
WL7 Methylobacterium 431 6059272 36478 6904
WL8 Methylobacterium 225 5322599 56100 6969
WL9 Methylobacterium 221 4670253 42885 6741
WL12 Methylobacterium 271 5545758 53038 694
WL18 Methylobacterium 1373 5796709 7880 6864
WL19 Methylobacterium 330 5177712 31633 6698
WL30 Methylobacterium 377 5636306 38077 6978
WL45 Roseomonas 2056 5961800 5511 6954
WL64 Methylobacterium 365 6908402 48840 6834
WL69 Methylobacterium 225 4693485 43639 6951
WL93 Methylobacterium 719 5642324 16902 6945
WL103 Methylobacterium 1327 5167285 7154 6897
WL116 Methylobacterium 1173 5445855 8413 6943
WL119 Methylobacterium 263 5709790 45485 6958
WL120 Methylobacterium 395 5266148 26698 6954
WL122 Methylobacterium 2024 4508072 3666 6877
Zervas et al GBE
2898 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Molecular Evolution of the PGC
The PGCs identified in the sequenced isolates all belong to
RC-II type meaning they contain bacteriochlorophyll-a and
possess pheophytinmdashquinone type reaction centers All
strains possess a full bacteriochlorophyll synthase gene set
except for Alsobacter sp WL4 that is lacking the bchM
gene (table 2)
For the carotenoid biosynthesis genes all strains possess
crtB whereas crtA is only present in Alsobacter sp WL4 and
Rhizobium sp strain WL3 For the crtC and crtK genes an
interesting pattern is observed Methylobacterium group 1
and group 2 strains contain crtC (with three exceptions dis-
cussed below) whereas group 3 and 4 strains completely lack
the gene On the other hand crtK is only present in
FIG 1mdashPhylogenetic tree of the 16S genes of the sequenced isolates Methylosinus trichosporium and Alsobacter metallidurans were included to better
pinpoint the phylogenetic placement of strain WL4 The legend shows substitution rates from the root of the tree and the node labels indicate bootstrap
support values ()
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2899
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Methylobacterium group 3 strains and missing from groups 1
2 and 4 Similar patterns are evidenced in pufB which is ab-
sent from Methylobacterium groups 3 and 4 Rhizobium sp
WL3 and Roseomonas sp WL45 and pufC missing from
Methylobacterium group 4 as well as the novel strain WL4
The remaining genes were found ubiquitously in all strains of
the analysis with only a few exceptions (table 2)
The PGCs were consistently found on the longer contigs of
the initial assemblies of Illumina data produced by SPAdes
(for the strains with fewer than 500 contigs) (table 1) In
Rhizobium sp WL3 and Roseomonas sp WL45 the photosyn-
thetic genes form one continuous cluster with different
however architectures In Alsobacter sp WL4 the genes are
organized in two clusters similar to the Methylobacterium
spp isolates In these cases the genes were found in the
middle of the contigs ruling out the possibility of PGC being
organized into two clusters as an artifact of the assembly
Unique to the architecture present in WL4 is the presence
FIG 2mdashPhylogenomic tree based on the core proteome (CheckM) of the 21 pure bacterial isolates of the analysis including Methylosinus trichosporium
OB3B and Alsobacter sp SH9 The legend shows substitution rates from the root of the tree and the node labels indicated bootstrap support values ()
Zervas et al GBE
2900 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
of bchI before the crtB crtI genes in the first cluster For
Methylobacterium spp groups 2 and 3 show similar gene
syntenymdashthough there are a few differences in gene content
(table 2)mdashwhereas groups 1 and 4 appear to also share the
overall architecture of the PGCs For isolates WL9 and WL19
(both in group 1) the exact location of the crtB and crtI genes
is not known as they appear on a short contig by themselves
Annotating the assemblies on RAST revealed several house-
keeping genes present in the vicinity of these clusters which
further supported the notion of their chromosomal place-
ment The completed genomes resulting from the hybrid as-
semblies (strains WL1 WL3 WL4 and WL45) provided the
final evidence of the exact placement of the PGCs on
the chromosomes of all our isolates The different
architectures of the isolated PGCs are shown in figure 5
General descriptive statistics of the assembled complete
genomes are given in table 3 The hybrid assemblies resulted
in both complete genomes and plasmids Thus Rhizobium sp
WL3 contains two plasmids Roseomonas sp WL45 contains
two chromosomes and seven plasmids Alsobacter sp WL4
with two plasmids Methylobacterium sp WL1 with one
plasmid
Using Roseomonas sp WL45 as outgroup we generated
phylogenetic trees for the 16S acsF bchL bchY crtB pufL
and pufM genes as described previously (data not shown)
We calculated substitution rates from the root of the different
trees including the core proteome (CheckM) tree and com-
pared them to each other (table 4)
Discussion
A New Strategy to Generate Complete Genomes ofUnique Aerobic Anoxygenic Phototrophic Bacteria FromEnvironmental Isolates
Aiming to enhance our knowledge of the prevalence of aer-
obic anoxygenic phototrophic bacteria in the phyllosphere
we designed this study using wheat as the plant of choice
and employed a high-throughput multi-disciplinary approach
to identify isolate and analyze AAP strains From the initial
137 plates we ended up with 4480 bacterial colonies
gt500 of which were probable AAP strains After macroscopic
investigation of size shape color and texture of the colonies
we chose 129 strains for which we ran proteome profiling
using MALDI-TOF MS This approach has been shown to be
FIG 3mdashPairwise proteome comparisons (BLAST-matrix) of the 21 bacterial isolates of the analysis The gray-to-green color gradient indicates low to
high percentages of identity The bottom boxes show the presence of homologs within the same proteome (gray-to-red color gradient shows low to high
presence of homologs in the specific proteome)
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2901
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
able to reproducibly distinguish different taxa and in cases
where reference MALDI-TOF spectra are available in
the database be able to assign Genus andor Species names
(Madsen et al 2015) We then proceeded with whole ge-
nome sequencing on the Illumina Nextseq platform and for
the unique genomes we also proceeded with Oxford
Nanopore Technologies sequencing to generate complete
genomes and plasmids
Following this multi-disciplinary approach that relies on
classical microbiological techniques (plating morphology
etc) biochemical analyses (MALDI-TOF MS) and two differ-
ent types of high-throughput whole genome sequencing we
were able to generate draft and complete genomes of novel
AAP bacteria from the wheat phyllosphere in a relatively quick
(weeks) and efficient manner without losing novel biological
information in the process of reducing redundancy
Bacteriochlorophyll-Containing AAP Bacteria in the WheatPhyllosphere
By investigating the presence of bacteriochlorophylls in the
wheat phyllosphere we were able to identify 21 unique strains
harboring them the majority of which belonged to the genus
Methylobacterium Members of this genus are ubiquitous in
nature and have been detected in both soil and plant surfaces
(Green 2006) Targeting the bchY (chlorophyllide reductase
subunit Y) and pufM (reaction centre subunit M) genes in
phyllosphere samples of different plants it was shown that
the Methylobacteriaceae (alpha-Proteobacteria) comprised
28 of the total pufM sequences from the amplicon sequenc-
ing analysis (Atamna-Ismaeel and Finkel 2012a) Our results
are thus in accordance to these previous studies As
Methylobacteriaceae are indigenous inhabitants of the phyllo-
sphere and have consistently been shown to harbor PGCs it
may be that these photosystems are essential to the
Methylobacteria that inhabit this niche Interestingly in their
study Atamna-Ismaeel et al had negative results from their
bchY amplicon sequencing approach suggesting the absence
of bacteria that contain RC1 (bacteriochlorophyll-a containing
reaction centers) type photosystems Our results show that all
isolates contain the complete cassette of chlorophyllide syn-
thase subunit encoding genes (bchB bchC bchF bchG bchI
bchL bchM bchN bchP bchX bchY bchZ) as well as both the
pufL and pufM genes (table 2) Testing the same primers (Yutin
et al 2009) in silico against our bchY sequences we had pos-
itive results for all of them (data not shown)
Patterns of loss of the carotenoid genes have been ob-
served in other alpha and gamma proteobacteria like in
Rhodobacterales and Sphingomonadales (Zheng et al
2011) The absence of crtA results in the inability to perform
the final step in the spheroidene pathway where hydroxy-
spheroidene is converted to hydroxyspheroidenone
Moreover the absence of crtC in Methylobacterium spp of
groups 3 and 4 completely affects both the spheroidene and
the spirilloxanthin pathways as its product is essential for the
first step of both metabolic pathways starting from neuro-
sporene and lycopene respectively This means that strains
FIG 4mdashWhole genome average nucleotide identity (ANI) cladogram
showing percentages of k-mer identities between the isolates Taxa with
the same color and above 90 identity threshold are considered to belong
in the same species Beijerinckia indica ATCC-9039 was used as outgroup
of the analyses (not defined a priori)
Zervas et al GBE
2902 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Tab
le2
Gen
eC
onte
nt
of
the
Photo
synth
esis
Gen
eC
lust
ers
Iden
tified
inth
e21
Bac
terial
Isola
tes
Bact
eri
och
loro
ph
yll
Syn
thase
Gen
es
Caro
ten
oid
Bio
syn
thesi
sG
en
es
Lig
ht-
Harv
est
ing
Co
mp
lex
Gen
es
Str
ain
acs
Fb
chB
bch
Cb
chF
bch
Gb
chI
bch
Lb
chM
bch
Nb
chP
bch
Xb
chY
bch
Zcr
tAcr
tBcr
tCcr
tDcr
tEcr
tFcr
tIcr
tKp
ucC
pu
fAp
ufB
pu
fCp
ufL
pu
fMp
uh
A
Rh
izo
biu
msp
W
L3
Ro
seo
mo
nas
sp
WL4
5
Als
ob
act
er
sp
WL4
Gro
up
1M
eth
ylo
bact
eri
um
sp
WL9
Meth
ylo
bact
eri
um
sp
WL1
9
Meth
ylo
bact
eri
um
sp
WL6
9
Gro
up
2M
eth
ylo
bact
eri
um
sp
WL1
Meth
ylo
bact
eri
um
sp
WL2
Meth
ylo
bact
eri
um
sp
WL7
Meth
ylo
bact
eri
um
sp
WL1
8
Meth
ylo
bact
eri
um
sp
WL6
4
Gro
up
3M
eth
ylo
bact
eri
um
sp
WL6
Meth
ylo
bact
eri
um
sp
WL3
0
Meth
ylo
bact
eri
um
sp
WL9
3
Meth
ylo
bact
eri
um
sp
WL1
16
Meth
ylo
bact
eri
um
sp
WL1
19
Gro
up
4M
eth
ylo
bact
eri
um
sp
WL8
Meth
ylo
bact
eri
um
sp
WL1
2
Meth
ylo
bact
eri
um
sp
WL1
03
Meth
ylo
bact
eri
um
sp
WL1
20
Meth
ylo
bact
eri
um
sp
WL1
22
NO
TEmdash
Gra
yb
oxe
sin
dic
ate
pre
sen
ceo
fg
en
es
an
dw
hit
eb
oxe
sin
dic
ate
ab
sen
ceo
fg
en
es
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2903
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
of these two groups rely on the zeaxanthin pathway to pro-
duce carotenoid pigments and most likely nostoxanthin and
erythroxanthin which however do not participate in the
light-harvesting process but rather have a photoprotection
role (Noguchi et al 1992 Yurkov and Csotonyi 2009)
Phylogenetic Observations
In our study we sequenced and analyzed 21 aerobic anoxy-
genic photosynthetic bacteria most of which belonged to the
genus Methylobacterium the presence of which in the
environment has already been discussed previously
Methylobacterium spp are facultative methylotrophs and
can utilize a variety of C1 C2 C3 and C4 compounds
Such compounds are often found in the phyllosphere due
to plant metabolism (Vorholt 2012 Remus-Emsermann
et al 2014) Methylobacterium spp show a wide range in
chromosome size and number of plasmids they contain
(Marx et al 2012) The genus is represented well in
GenBank with 100 complete or draft genomes (last accessed
April 2019) However most of the sequences do not have
FIG 5mdashArchitecture of the different photosynthetic gene clusters Arrows indicate the orientation of the genes Sizes are not representatives of gene
length Only the genes shared by all PGCs were shown Genes that may be missing from some strains are denoted with parentheses on the figurersquos legend
Table 3
General Descriptive Statistics of the Four Complete AAP Bacterial Genomes
Taxon Genome Size (bp) GC Number of Genes Plasmids Plasmids Size (bp) GC
Rhizobium sp WL3 4568855 6160 4450 1 700631 586
mdash mdash mdash 2 85350 575
Roseomonas sp WL45 4533887 708 5042 1 262815 655
1011854 707 1138 2 240556 662
mdash mdash mdash 3 152922 658
mdash mdash mdash 4 85279 646
mdash mdash mdash 5 84774 661
mdash mdash mdash 6 83273 656
mdash mdash mdash 7 65860 648
Alsobacter sp WL4 5405668 679 5013 1 27178 620
mdash mdash mdash 2 9451 606
Methylobacterium sp WL1 6215463 691 6431 1 35731 638
Zervas et al GBE
2904 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
species affiliation (fig 4) Thus we were not able to assign any
of the 18 isolates to specific Methylobacterium species apart
perhaps from group 2 (WL1 WL2 WL7 WL18 WL64) which
is placed as sister to M brachiatum in the MASH-ANI analysis
Even in this case though M pseudosasicola BL36 is also
placed in the same clade as M brachiatum which raises the
question of proper nomenclature This is further evidenced in
the same tree (cyanide clade below) where seven species
share gt90 of their genome For the remaining isolates
WL19 is closely affiliated to other Methylobacterium strains
that have not been identified A similar picture is shown for
WL69 On the other hand WL19 appears to be a novel spe-
cies with unique genomic features and so do the strains from
groups 3 and 4 (red clade) Thus despite how ubiquitous
Methylobacterium is in nature it is clear that part of its diver-
gence is still missing Moreover a critical evaluation of the
current nomenclature of the genus would be beneficial based
on the whole genome sequences because closely related
genomes (MASH-ANI identitygt 97) have been given differ-
ent species affiliations
Apart from this genus we also identified a Rhizobium sp
a Roseomonas sp and a novel Alsobacter strain (WL4) that
belongs in the Methylocystaceae a family of methanotroph
bacteria This is the first time that AAP bacteria from this
family have been isolated and their complete genomes assem-
bled Its closely related Alsobacter spp have been shown to
exhibit high Thallium tolerance (Bao et al 2006) and it will be
interesting to identify heavy-metal resistance genes in non-
heavy-metal polluted phyllosphere samples It is worth men-
tioning that if WL4 contains heavy-metal resistance genes on
top of the PGC this will have been shown for the first time
and its potential to bioremediate heavy-metal contaminated
areas if applied as plant growth promotion and environmental
remediation agent is of high value and needs to further be
explored
Molecular Evolution of the PGCs
For all gene trees of the PGC we observed that Rhizobium sp
WL3 evolves faster than the other isolates followed by
Table 4
Substitutions Per Site in the Different Phylogenetic Analyses
Strain 16S acsF bchL bchY crtB pufL pufMsuper-matrix
CheckM
Roseomonas sp WL45 0096 0147 0100 0235 0250 0095 0132 0160 0184
Rhizobium sp WL3 0184 0622 0767 0690 1370 0343 0525 0580 0372
Alsobacter sp WL4 0151 0462 0233 0490 0870 0258 0334 0380 0403
Methylobacterium sp WL9 0226 0418 0372 0420 0670 0270 0319 0364 0477
Methylobacterium sp WL19 0221 0413 0300 0460 0600 0233 0346 0364 0468
Methylobacterium sp WL69 0231 0373 0333 0470 0620 0260 0358 0368 0468
Methylobacterium sp WL1 0233 0422 0344 0485 0480 0280 0385 0380 0459
Methylobacterium sp WL2 0233 0422 0344 0485 0480 0280 0385 0380 0459
Methylobacterium sp WL7 0238 0427 0344 0480 0480 0278 0385 0380 0459
Methylobacterium sp WL18 0238 0427 0344 0480 0480 0280 0385 0380 0459
Methylobacterium sp WL64 0233 0422 0344 0480 0485 0280 0385 0380 0459
Methylobacterium sp WL6 0199 0378 0256 0425 0680 0208 0311 0340 0455
Methylobacterium sp WL30 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL93 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL116 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL119 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL8 0188 0373 0258 0430 0680 0230 0358 0348 0459
Methylobacterium sp WL12 0188 0373 0256 0430 0680 0230 0358 0348 0459
Methylobacterium sp WL103 0188 0373 0256 0430 0850 0230 0358 0364 0459
Methylobacterium sp WL120 0188 0373 0258 0430 0680 0233 0358 0348 0459
Methylobacterium sp WL122 0188 0373 0256 0430 0680 0230 0358 0348 0459
Susbstitutions per site
Gro
up1
Gro
up 2
Gro
up 3
Gro
up 4
NOTEmdashAll values correspond to distances from the root where Roseomonas sp WL45 was chosen as outgroup for all the alignments
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2905
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Alsobacter sp WL4 Both these isolates also show much
higher substitution rates compared with their respective 16S
genes and core proteomes (checkM) which indicates that the
PGCs are under relaxed selection pressure In all instances the
phylogenetic trees constructed showed the same topology
and the same grouping of the Methylobacterium isolates
which suggests that the PGCs were incorporated in the
genomes of these strains relatively early in their evolution If
the PGC was free to move laterally then we would have ob-
served very different phylogenetic trees for its genes com-
pared with the 16S and core proteome analyses
Interestingly these observations do not reflect on the archi-
tecture of the PGC where Roseomonas sp WL45 and
Rhizobium sp WL3 contain the PGC in one continuous re-
gion whereas Alsobacter sp WL4 and Methylobacterium sp
WL1 have it in two pieces
For the 18 Methylobacterium isolates group 1 and 2
strains (WL9 WL19 WL69 and WL1 WL2 WL7 WL18
WL64) evolve consistently at different rates compared
with the groups 3 and 4 It appears that there is relatively
strong selection pressure on bchL pufL and pufM which
in all cases are more similar to each other and also show
slower molecular evolution rates compared with ascF
bchY and crtB which in all Methylobacterium isolates
evolve at least twice as fast compared with the 16S
gene (table 4) They also show fewer substitutions per
site compared with the core proteome analysis This result
partially contrasted with previously adopted methods of
using bchY as a marker gene for RC1 clusters (eg
Atamna-Ismaeel and Finkel 2012a) On the other hand
the use of pufM appears more sensible It would how-
ever be more suitable to investigate the possibility of us-
ing other genes for metagenome amplicon sequencing
approaches such as pufL or bchL which show a smaller
degree of divergence in different genera thus universal
primers might be more successful in detecting a wide va-
riety of AAP bacteria in environmental samples
Interestingly while Groups 1 and 2 isolates exhibit higher
substitution rates of the PGC genes compared with
Groups 3 and 4 this is not the case for crtB which evolves
a lot slower and also pufM which shows similar molec-
ular evolution in the genus Overall it seems that the
PGCs were incorporated in ancestral strains that later di-
verged to give rise to the different AAP taxa This is fur-
ther corroborated by the fact that almost all phylogenetic
trees (eg fig 5) are consistent and in agreement with
both the 16S phylogenetic tree (fig 1) as well as the phy-
logenomic tree (fig 2) If the genes of the PGCs had been
transferred to these AAP several times there would have
been significant differences both in the sequence level
and the phylogenetic analysis of the different genes
which would be expected to result in different topologies
and diverse distances from the root even for closelymdash
based on 16S analysismdashrelated strains
The Significance of AAP Bacteria in Wheat Phyllosphere
In this study we isolated for the first time a variety of AAP
bacteria that are present in the wheat phyllosphere adding to
the metagenomics knowledge provided earlier from five other
plants (Atamna-Ismaeel and Finkel 2012a 2012b) The phyllo-
sphere is a rather harsh environment with little water avail-
ability scarce resources very few suitable locations extreme
temperature difference between day and night and high
dosages of UV radiation (Vorholt 2012) AAP bacteria have
the ability to utilize light in photophosphorylation processes
through which they produce ATP and at the same time funnel
the few available nutrients in other metabolic pathways
(Yurkov and Hughes 2017) This has been shown to give
them an edge in vitro after exposure to light compared with
nonAAP bacteria (Ferrera et al 2011) Considering that phyl-
losphere bacteria protect the plant from pathogenic microbes
by secreting metabolites and by occupying the available
niches being able to persevere and outgrow other bacteria
gives a significant edge to AAP taxa and probably points
toward the suggestion of positive selection of such bacteria
in their phyllosphere by the plants themselves as they appear
to be absent from the soil (Atamna-Ismaeel and Finkel
2012b) At this point it is not clear exactly how AAP bacteria
benefit the plant especially since the phyllosphere is a com-
plex habitat with an ever-changing harsh environment In the
past years great efforts have been carried out on how to
improve crop production by studying the rhizosphere It has
also been suggested that the phyllosphere communities play a
significant role in plant health and growth which ultimately
leads to increased yields (Lindow and Brandl 2003) Given the
current projections of world population growth and the sup-
ply of food (UN 2017) it becomes evident that more in-depth
studies investigating the phyllosphere of plants and especially
crops and identifying the underlying drivers of the bacterial
communities and their interactions with their plant hosts is of
paramount importance Thus by isolating maintaining and
analyzing the genomes of phyllosphere isolates we may iden-
tify strains that may prove useful as plant growth promoting
agents in future field trials investigating the effect of phyllo-
sphere inoculation on plant growth health and yield
Supplementary Material
Supplementary data are available at Genome Biology and
Evolution online
Acknowledgments
The authors would like to thank Tue Sparholt Jorgensen for
collecting the wheat sample The authors also want to thank
Tina Thane (AU) Tanja Begovic (AU) and Margit Frederiksen
(NRCWE) for capable laboratory assistance We thank Michal
Koblızek for providing the initial model of the colony infrared
imaging system This work was funded by a Villum
Zervas et al GBE
2906 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Experiment grant (no 17601) and a Marie Skłodowska-Curie
AIAS-COFUND fellowship (EU-FP7 program Grant
Agreement No 609033) awarded to YZ
Author Contributions
AZ YZ LHH designed the study YZ isolated the strains
performed colony infrared imaging and extracted DNA for
Illumina sequencing YZ and AMM performed the MALDI-
TOF MS analysis AZ extracted DNA and performed ONT
Sequencing AZ carried out the bioinformatics and phyloge-
netic analyses AZ and YZ drafted the manuscript AZ
YZ AMM and LHH revised the manuscript for
intellectual content All authors read and approved the
manuscript
Literature Cited
Atamna-Ismaeel N Finkel O 2012a Bacterial anoxygenic photosynthesis
on plant leaf surfaces Environ Microbiol Rep 4(2)209ndash216
Atamna-Ismaeel N Finkel OM 2012b Microbial rhodopsins on leaf sur-
faces of terrestrial plants Environ Microbiol 14(1)140ndash146
Aziz RK et al 2008 The RAST Server rapid annotations using subsystems
technology BMC Genomics 9(1)75
Bankevich A et al 2012 SPAdes a new genome assembly algorithm and
its applications to single-cell sequencing J Comput Biol
19(5)455ndash477
Bao Z Sato Y Kubota M Ohta H 2006 Isolation and characterization of
thallium-tolerant bacteria from heavy metal-polluted river sediment
and non-polluted soils Microb Environ 21(4)251ndash260
Beatie GA 2002 Leaf surface waxes and the process of leaf colonization
by microorganisms In Lindow SE Hecht-Poinar EI Elliott VJ editors
Phyllosphere Microbiology St Paul (USA) APS Press pp 3ndash26
Beattie GA Lindow SE 1999 Bacterial colonization of leaves a spectrum
of strategies Phytopathology 89(5)353ndash359
Brettin T et al 2015 RASTtk a modular and extensible implementation of
the RAST algorithm for building custom annotation pipelines and an-
notating batches of genomes Sci Rep 58365
Carvalho SD Castillo JA 2018 Influence of light on plantndashphyllosphere
interaction Front Plant Sci 91482
Ferrera I Gasol JM Sebastian M Hojerova E Koblızek M 2011
Comparison of growth rates of aerobic anoxygenic phototrophic bac-
teria and other bacterioplankton groups in coastal Mediterranean wa-
ters Appl Environ Microbiol 77(21)7451ndash7458
Ferrera I Sanchez O Kolarova E Koblızek M Gasol JM 2017 Light
enhances the growth rates of natural populations of aerobic anoxy-
genic phototrophic bacteria ISME J 11(10)2391ndash2393
Gorlach K Shingaki R Morisaki H Hattori T 1994 Construction of eco-
collection of paddy field soil bacteria for population analysis J Gen
Appl Microbiol 40(6)509ndash517
Gourion B Rossignol M Vorholt JA Lindow SE 2006 A proteomic study
of Methylobacterium extorquens reveals a response regulator essential
for epiphytic growth wwwintegratedgenomicscom Accessed
November 26 2018
Green PN 2006 Methylobacterium In Dworkin M Falkow S Rosenberg
E Schleifer K Stackebrandt E editors The prokaryotes ndash a handbook
on the biology of bacteria New York Springer p 257ndash265
Hauruseu D Koblızek M 2012 Influence of light on carbon utilization in
aerobic anoxygenic phototrophs Appl Environ Microbiol
78(20)7414ndash7419
Katoh K Misawa K Kuma K Miyata T 2002 MAFFT a novel method for
rapid multiple sequence alignment based on fast Fourier transform
Nucleic Acids Res 30(14)3059ndash3066
Knief C et al 2012 Metaproteogenomic analysis of microbial communi-
ties in the phyllosphere and rhizosphere of rice ISME J
6(7)1378ndash1390
Knief C Frances L Vorholt JA 2010 Competitiveness of diverse methyl-
obacterium strains in the phyllosphere of arabidopsis thaliana and
identification of representative models including M extorquens
PA1 Microb Ecol 60(2)440ndash452
Kwak MJ et al 2014 Genome information of Methylobacterium oryzae a
plantndashprobiotic methylotroph in the phyllosphere PLoS One
9(9)e106704
Lindow SE Brandl MT 2003 Microbiology of the phyllosphere Appl
Environ Microbiol 69(4)1875ndash1883
Madsen AM Zervas A Tendal K Nielsen JL 2015 Microbial diversity in
bioaerosol samples causing ODTS compared to reference bioaerosol
samples as measured using Illumina sequencing and MALDI-TOF
Environ Res 140255ndash267
Martin M 2011 Cutadapt removes adapter sequences from high-
throughput sequencing reads EMBnet J 17(1)10ndash12
Marx CJ et al 2012 Complete genome sequences of six strains of the
genus Methylobacterium b J Bacteriol 194(17)4746
Noguchi T Hayashi H Shimada K Takaichi S Tasumi M 1992 In vivo
states and functions of carotenoids in an aerobic photosynthetic bac-
terium Erythrobacter longus Photosynth Res 31(1)21ndash30
Olm MR Brown CT Brooks B Banfield JF 2017 dRep a tool for fast and
accurate genomic comparisons that enables improved genome recov-
ery from metagenomes through de-replication ISME J
11(12)2864ndash2868
Overbeek R et al 2014 The SEED and the Rapid Annotation of microbial
genomes using Subsystems Technology (RAST) Nucl Acids Res
42(D1)D206ndashD214
Parks DH Imelfort M Skennerton CT Hugenholtz P Tyson GW 2015
CheckM assessing the quality of microbial genomes recovered from
isolates single cells and metagenomes Genome Res 25(7)1043ndash1055
Remus-Emsermann MNP et al 2014 Spatial distribution analyses of nat-
ural phyllosphere-colonizing bacteria on Arabidopsis thaliana revealed
by fluorescence in situ hybridization Environ Microbiol
16(7)2329ndash2340
Stamatakis A 2006 RAxML-VI-HPC maximum likelihood-based phyloge-
netic analyses with thousands of taxa and mixed models
Bioinformatics 22(21)2688ndash2690
UN 2017 World population prospects the 2017 revision Data Booklet
httpspopulationunorgwppPublicationsFilesWPP2019_DataBook-
letpdf
Vesth T Lagesen K Acar euroO Ussery D 2013 CMG-biotools a free work-
bench for basic comparative microbial genomics PLoS One
8(4)e60120
Vorholt JA 2012 Microbial life in the phyllosphere Nat Rev Microbiol
10(12)828ndash840
Wick RR Judd LM Gorrie CL Holt KE 2017 Unicycler resolving bacterial
genome assemblies from short and long sequencing reads PLoS
Comput Biol 13(6)e1005595
Wilson M Lindow SE 1994a Coexistence among epiphytic bacterial pop-
ulations mediated through nutritional resource partitioning Appl
Environ Microbiol 604468ndash77
Wilson M Lindow SE 1994b Ecological similarity and coexistence of epi-
phytic ice-nucleating (ice) pseudomonas syringae strains and a non-
ice-nucleating (ice) biological control agent Appl Environ Microbiol
60(9)3128ndash3137
Woodward FI Lomas MR 2004 Vegetation dynamics ndash simulating
responses to climatic change Biol Rev 79(3)643ndash670
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2907
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Xu J Gordon JI 2003 Honor thy symbionts Proc Natl Acad Sci USA
100(18)10452ndash10459
Yurkov V Csotonyi JT 2009 New light on aerobic anoxygenic photo-
trophs In Hunter NC Daldal F Thurnauer MC Beatty JT editors
The purple photosynthetic bacteria Dordrecht Springer thorn Business
Media BV p 31ndash55 doi101007978-1-4020-8815-5_3
Yurkov V Hughes E 2017 Aerobic Anoxygenic Phototrophs four
decades of mystery In Hallenbeck PC editor Modern topics in the
phototrophic prokaryotes environmental and applied aspects p
193ndash214
Yutin N et al 2009 BchY-based degenerate primers target all types of
anoxygenic photosynthetic bacteria in a single PCR Appl Environ
Microbiol 75(23)7556ndash7559
Zeng Y Feng F Medova H Dean J Koblizek M 2014 Functional
type 2 photosynthetic reaction centers found in the rare bacterial phy-
lum Gemmatimonadetes Proc Natl Acad Sci USA 111(21)7795ndash7800
Zheng Q et al 2011 Diverse arrangement of photosynthetic gene
clusters in aerobic anoxygenic phototrophic bacteria PLoS One
6(9)e25050
Associate editor Esperanza Martinez-Romero
Zervas et al GBE
2908 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
- evz204-TF1
- evz204-TF2
-
by microorganisms of the phyllosphere However the most
significant environmental driver is sunlight Temperature dif-
ferences in the upper leaves especially those that form the
canopy of various habitats may range up to 50 C between
day and night Sunlight though beneficial for photosynthetic
organisms is also comprised of UV-radiation which is dam-
aging to the DNA of both prokaryotes and eukaryotes
(Remus-Emsermann et al 2014) In addition to the abiotic
factors that make the phyllosphere a much more hostile en-
vironment than the rhizosphere its inhabitants also facemdash
directly or indirectlymdashthe influence of their plant host
Leaves are surrounded by a thin laminar and waxy layer
the cuticle which renders the leaf surface hydrophobic
thus removing the excess of water that is collected due to
rainfall dew or respiration from the stomata This leads to the
retention of little water most usually in veins and other cavities
of the cuticle These microformations also protect the colo-
nizers from the surrounding environment (Beatie 2002) Apart
from niche competition and scarce water availability phyllo-
sphere colonizers also need to compete for nutrients as the
cuticle makes the surface virtually impermeable to nutrients
deriving from diffusion through the cells of the plant host
(Wilson and Lindow 1994a 1994b) At the same time they
also need to protect themselves from potential invaders and a
plethora of antimicrobial compounds of prokaryotic or eu-
karyotic origin (Vorholt 2012)
Albeit the harsh conditions they encounter phyllo-
sphere microogransims exhibit remarkable biodiversity
which in certain cases can be comparable to that of hu-
man gut microbiome (Xu and Gordon 2003 Knief et al
2012) Through recent genomics and metagenomics stud-
ies several bacterial genera such as Methylobacterium
have been shown to be ubiquitous in the phyllosphere
(Knief et al 2010) and their role in nutrient recycling plant
growth promotion and protection has been evidenced pre-
viously (Beattie and Lindow 1999 Gourion et al 2006
Atamna-Ismaeel and Finkel 2012a Kwak et al 2014)
One of the most interesting recent findings in phyllosphere
microbiota was the presence of aerobic anoxygenic photo-
trophic (AAP) bacteria and rhodopsin-harboring bacteria in
a variety of land plants (Atamna-Ismaeel and Finkel 2012a
2012b) AAP bacteria are commonly found in aquatic envi-
ronments ranging from the arctic to the tropics and from
high salinity lakes to pristine high altitude lakes (Yurkov and
Hughes 2017) They rely on organic carbon compounds to
cover their nutritional requirements and can also utilize
light to produce energy in the form of ATP without fixing
carbon or producing oxygen AAPs have been found in a
variety of extreme environments and have shown their po-
tential to outgrow nonAAP bacteria under nutrient limiting
conditions (Hauruseu and Koblızek 2012 Ferrera et al
2017) Thus their presence in the phyllosphere may not
be surprising However their relative abundance at least
in the rice phyllosphere (Oryza sativa) was up to three
times higher than what is commonly found in marine eco-
systems (Atamna-Ismaeel and Finkel 2012a)
These early culture-independent metagenomics studies
provided valuable information about the presence of AAP
bacteria in the phyllosphere However no pure cultures of
AAPs have been described so far from phyllosphere and
thus detailed genome information is lacking which prevents
an in-depth understanding of the ecological roles and evolu-
tionary trajectory of AAPs in this unique niche To expand our
genomics views of this ecologically important group of bac-
teria in the phyllosphere we designed this study with the
following aims 1) to create the first collection of AAP isolates
from the phyllosphere 2) to characterize their photosynthesis
gene clusters (PGCs) and compare their gene contents and
molecular evolutionary patterns and 3) to expand the data-
base of complete genomes of AAP bacteria from nonaquatic
environments By combining colony infrared (IR) imaging
MALDI-TOF mass spectrometry (MS) and Illumina and
Oxford Nanopore sequencing technologies we were able to
provide 20 draft genomes that contain complete PGCs and
four complete genomes that contain plasmids The further
comparison of the evolutionary trajectories of their PGCs
revealed a diverging pattern among the highly homogenous
Methylobacterium strains highlighting an ongoing genomic
innovation at the gene cluster level
Materials and Methods
Sample Collection
A whole intact plant of winter wheat (Triticum aestivum L)
with a height of circa 60 cm was collected from a field in
Roskilde Denmark on June 6 2018 and transported to the
lab for bacterial isolation on the same day Wheat leaves were
cut off and cleaned with running sterile water to remove dust
and other temporary deposits from the ambient environment
Microbiota were collected from eight pieces of leaves by rins-
ing the leaves in 80 ml PBS solution (pH 74) in a Oslash 150-mm
petri dish and scraping the leave surface repeatedly with ster-
ile swabs until the PBS solution slightly turned greenish due to
the fall-off of leave cells From the supernatant 14 ml were
plated on 15 strength R2A plates (resulting in a total of 137
agar plates) and incubated at room temperature with normal
indoor light conditions for two weeks AAP bacteria were
identified using a custom screening chamber equipped with
an IR imaging system described in detail previously (Zeng
et al 2014) In short plates were placed in a chamber sealed
from light and exposed to a source of green lights AAP bac-
teria containing bacteriochlorophyll absorb the green light
and emit radiation in the near IR spectrum which was cap-
tured by a NIR sensitive CMOS camera The image was proc-
essed and the glowing colonies indicated the presence of
BChl a The colonies that give positive signals were marked
restreaked onto 15 strength R2A plates and incubated at
room temperature for 72 h Verification of the light absorbing
Zervas et al GBE
2896 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
capabilities of the isolated strains was performed using the
same setup Strains that passed the verification step were
further restreaked onto 15 strength R2A plates to ensure
pure colonies of single AAP strains
Selection of AAP Strains and Whole Genome Sequencing
The isolated AAP strains were grouped and dereplicated using
a MALDI-TOF MS (Microflex LT Bruker Daltonics Bremen
Germany) before performing whole genome sequencing to
reduce the cost while maintaining the biodiversity Briefly a
toothpick was used to transfer a small amount of an AAP
bacterial colony onto the target plate (MSP 96 polished steel
Bruker) that was evenly spread out and formed a thin layer of
biomass on the steel plate The sample was then overlaid with
70 formic acid and allowed for air dry before addition of
1lL MALDI-MS matrix solution (a-cyano-4-hydroxycinnamic
acid SigmandashAldrich) A bacterial standard with well-
characterized peaks (Bruker Daltonics) was used to calibrate
the instrument The standard method ldquoMBT_AutoXrdquo was ap-
plied to obtain proteome profiles within the mass range of 2ndash
20 kDa using the flexControl software (Bruker) The
flexAnalysis software (Bruker) was used to smooth the data
subtract baseline and generate main spectra (MSP) followed
by a hierarchical clustering analysis with the MALDI Biotyper
Compass Explorer software which produced a dendrogram
as the output for visual inspection of similarities between
samples An empirical distance value of 50 was used as the
cutoff for defining different groups on the dendrogram cor-
responding to different strainsspecies
From each group on the dendrogram one isolate was cho-
sen and restreaked on 1=2 R2A plates and incubated at room
temperature for one week Prior to DNA isolation the plates
were again tested for the presence of light absorbing pigments
as previously described Bacterial colonies were scraped from
the plates using a loop and immersed in 400mL PBS solution
vortexed until the cells had separated spun down in a table-top
centrifuge at 10000 g at 4 C the supernatant was removed
and the pellets were redissolved in milliQ H2O High molecular
weight DNA was extracted from each strain using a modified
Masterpure Complete DNA amp RNA Purification Kit (Lucigen) by
replacing the elution buffer with a 10mM TrisndashHClmdash50mM
NaCl (pH 75ndash80) solution at the final step DNA concentration
was quantified on a Qubit 20 (Life Technologies) using the
Broad Range DNA Assay kit and DNA quality was measured
on a Nanodrop 2000C (Thermo Scientific) Whole genome
shotgun sequencing was performed on all AAP strains on the
Illumina Nextseq platform in house using the 2150bp chem-
istry Selected strains were also sequenced on the MinION plat-
form (Oxford Nanopore Technologies UK) in house using the
Rapid Barcoding Kit (SBQ-RBK004) and the FLO-MIN106 flow
cell (R941) following the manufacturerrsquos instructional
manuals
Genome Assembly and Analyses
The Illumina pair end reads were trimmed for quality and
ambiguities using Cutadapt (Martin 2011) and assembled
using SPAdes (Bankevich et al 2012) The resulting assem-
blies were analyzed using dRep (Olm et al 2017) to compare
their average nucleotide identities (ANI) in order to identify
clusters of closely related taxa The assemblies were
imported in Geneious R1125 (Biomatters Ltd) and PGCs
were annotated using the Roseobacter litoralis strain
Och149 plasmid pRL0149 (CP002624) as reference under
relaxed similarity criteria (40) The suggested genes of
probable PGCs were analyzed using BLAST (megablast
BlastN) to verify their annotations For full genome annota-
tions the assemblies were uploaded on RAST (Aziz et al
2008 Overbeek et al 2014 Brettin et al 2015) The pre-
dicted translated protein coding genes were imported in
CMG-Biotools (Vesth et al 2013) for pairwise proteome
comparisons using the native blastmatrix program
Hybrid assemblies of pair-end Illumina reads and long
Oxford Nanopore Technologies reads were performed using
the Unicycler assembler (Wick et al 2017) utilizing the bold
assembly mode The resulting complete genomes were
imported in Geneious where their circularity was confirmed
by mapping-to-reference runs using the short and long reads
from the respective hybrid assemblies and the native
Geneious mapper under default settings in the Mediumndash
Low sensitivity mode
Phylogenetic Analyses
Identification of the 16S ribosomal DNA sequences was per-
formed by rnammer in CMG-Biotools Data sets containing
the different genes from the identified PGCs were created in
Geneious The resulting sequences were aligned using MAFFT
v7388 (Katoh et al 2002) and it FFT-NS-I x1000 algorithm
with default settings as implemented in Geneious
Phylogenetic trees were created using RAxML (Stamatakis
2006) employing the GTR-GAMMA nucleotide substitution
model and the rapid bootstrapping and search for best scor-
ing ML-tree with 100 bootstraps replicates and starting from
a random tree Analysis of the PGCs was performed on five
selected genes that were present in all isolates namely acsF
bchL bchY crtB pufL and pufM Individual gene alignments
and phylogenetic trees were performed as described above
A super-matrix (8240 characters) comprised of the total
seven gene alignments was created using the built-in
ldquoConcatenate alignmentsrdquo function in Geneious and a phy-
logenetic tree was created as above All trees were visually
inspected for disagreements Finally a phylogenomic tree
was constructed in RAxML employing the GAMMA
BLOSUM62 substitution model and the rapid bootstrapping
and search for best scoring ML-tree algorithm with 100 boot-
straps replicates and starting from a random tree using as
input the core protein alignment consisting of 6988
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2897
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
characters created with CheckM (Parks et al 2015) and its
lineage_wf algorithm with default settings Substitution rates
were calculated as distances from the root using the legend
provided by the phylogenetic trees
Results
Identification and Isolation of AAP Bacteria from thePhyllosphere
The 129 phyllosphere isolates that were tested positive for the
presence of bacteriochlorophyll-a were placed in 21 groups
according to their protein mass profiles based on MALDI-TOF
MS (supplementary fig MALDI-TOF Supplementary Material
online) A representative from each group was sequenced
(Illumina) and their genomes assembled (SPAdes) WL1 and
WL2 were from the same MALDI-TOF group and their
genomes are identical (ANI 100) confirming the validity
of MALDI-TOF MS for grouping similar isolates Information
about their phylogeny total genome size and genome as-
sembly statistics are presented in table 1
Analysis of the 16S genes showed that 17 isolates
belonged in Methylobacterium (separated in four distinct
groups) whereas the remaining three belonged to
Rhizobium Roseomonas and one novel species was affiliated
to the Methylocystaceae family (WL4) with most hits belong-
ing to uncultured environmental strains The only character-
ized hits of WL4 belonged to Methylosinus trichosporium
(Gorlach et al 1994) and Alsobacter metallidurans a recently
characterized novel genus novel species (Bao et al 2006)
However in both instances the similarity was low
(985) We included strains of both species and created
a 16S phylogenetic tree (fig 1) which placed Alsobacter sp
WL4 as sister to the Alsobacter-Methylosinus clade
The core proteome analysis (fig 2mdashCheckM) including M
trichosporium OB3b and Alsobacter sp SH9 established the
phylogeny of strain WL4 and it was highly supported as sister
to Alsobacter sp SH9 (BSfrac14 100) both of which were
distinct from M trichosporium OB3b (BSfrac14 100)
Methylobacterium strains were still placed into four distinct
groups (Group 1 WL9 WL19 WL69 Group 2 WL1 WL2
WL7 WL18 WL64 Group 3 WL6 WL30 WL93 WL116
WL119 Group 4 WL8 WL12 WL103 WL120) This was
further supported by the whole proteome comparison
(fig 3mdashBLAST matrix) For Groups 3 and 4 the average
core protein content is 4500 proteins whereas for Group
2 it is close to 4000 This may signify the presence of
either more or larger plasmids that characterize the strains
of this group For Group 1 this number falls even more as
in this case there clearly are three distinct species that make
up the group
We expanded the analysis to include 100 complete and
draft Methylobacterium spp genomes available in GenBank
(last assessed April 2019) and also other representatives from
the Methylocystaceae aiming at further establishing strain
WL4 in the genus Alsobacter and also providing greater res-
olution on the Methylobacterium isolates with a special focus
on the divergent group 1 Information about the included
genomes is given in supplementary table 1 Supplementary
Material online We performed whole genome comparisons
using ANI (fig 4mdashdRep)
The analysis showed that Group 1 isolates belonged in
different Methylobacterium species groups 3 and 4 were dif-
ferent strains of the same species (MASH ANI identity be-
tween groupsgt 90) whereas group 3 isolates are closely
related strains possibly belonging to Methylobacterium bra-
chiatum Strain WL4 is again placed close to the two identical
Alsobacter genomes (Alsobacter sp SH9 and A metallidurans
PVZS01) far away from the three remaining species from
within the Methylocystaceae (M trichosporium OB3b
Methylocystis rosea GW6 Methylocystis bryophila S285)
However the MASH-ANI identity between WL4 and
Alsobacter is below the 80 threshold of genus assignment
Therefore we took the top 200 megablast hits for WL4rsquos 16S
region (information provided in supplementary table 2
Supplementary Material online) aligned them including strain
WL4 as mentioned previously and created a phylogenetic tree
of the 201 sequences (supplementary fig 1 Supplementary
Material online) The results clearly show that WL4 belongs in
Alsobacter whereas Methylosinus spp form their own dis-
tinct cluster
Table 1
Genome Statistics and Phylogenetic Information of the 21 Isolates that
were Illumina Sequenced in This Study
Strain Taxonomy Contigs Genome Size N50 GC
WL1 Methylobacterium 654 6258106 19384 6892
WL2 Methylobacterium 298 6292310 52057 6899
WL3 Rhizobium 55 5333961 210875 6114
WL4 Alsobacter 139 5412272 86934 6792
WL6 Methylobacterium 708 5549874 15034 6972
WL7 Methylobacterium 431 6059272 36478 6904
WL8 Methylobacterium 225 5322599 56100 6969
WL9 Methylobacterium 221 4670253 42885 6741
WL12 Methylobacterium 271 5545758 53038 694
WL18 Methylobacterium 1373 5796709 7880 6864
WL19 Methylobacterium 330 5177712 31633 6698
WL30 Methylobacterium 377 5636306 38077 6978
WL45 Roseomonas 2056 5961800 5511 6954
WL64 Methylobacterium 365 6908402 48840 6834
WL69 Methylobacterium 225 4693485 43639 6951
WL93 Methylobacterium 719 5642324 16902 6945
WL103 Methylobacterium 1327 5167285 7154 6897
WL116 Methylobacterium 1173 5445855 8413 6943
WL119 Methylobacterium 263 5709790 45485 6958
WL120 Methylobacterium 395 5266148 26698 6954
WL122 Methylobacterium 2024 4508072 3666 6877
Zervas et al GBE
2898 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Molecular Evolution of the PGC
The PGCs identified in the sequenced isolates all belong to
RC-II type meaning they contain bacteriochlorophyll-a and
possess pheophytinmdashquinone type reaction centers All
strains possess a full bacteriochlorophyll synthase gene set
except for Alsobacter sp WL4 that is lacking the bchM
gene (table 2)
For the carotenoid biosynthesis genes all strains possess
crtB whereas crtA is only present in Alsobacter sp WL4 and
Rhizobium sp strain WL3 For the crtC and crtK genes an
interesting pattern is observed Methylobacterium group 1
and group 2 strains contain crtC (with three exceptions dis-
cussed below) whereas group 3 and 4 strains completely lack
the gene On the other hand crtK is only present in
FIG 1mdashPhylogenetic tree of the 16S genes of the sequenced isolates Methylosinus trichosporium and Alsobacter metallidurans were included to better
pinpoint the phylogenetic placement of strain WL4 The legend shows substitution rates from the root of the tree and the node labels indicate bootstrap
support values ()
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2899
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Methylobacterium group 3 strains and missing from groups 1
2 and 4 Similar patterns are evidenced in pufB which is ab-
sent from Methylobacterium groups 3 and 4 Rhizobium sp
WL3 and Roseomonas sp WL45 and pufC missing from
Methylobacterium group 4 as well as the novel strain WL4
The remaining genes were found ubiquitously in all strains of
the analysis with only a few exceptions (table 2)
The PGCs were consistently found on the longer contigs of
the initial assemblies of Illumina data produced by SPAdes
(for the strains with fewer than 500 contigs) (table 1) In
Rhizobium sp WL3 and Roseomonas sp WL45 the photosyn-
thetic genes form one continuous cluster with different
however architectures In Alsobacter sp WL4 the genes are
organized in two clusters similar to the Methylobacterium
spp isolates In these cases the genes were found in the
middle of the contigs ruling out the possibility of PGC being
organized into two clusters as an artifact of the assembly
Unique to the architecture present in WL4 is the presence
FIG 2mdashPhylogenomic tree based on the core proteome (CheckM) of the 21 pure bacterial isolates of the analysis including Methylosinus trichosporium
OB3B and Alsobacter sp SH9 The legend shows substitution rates from the root of the tree and the node labels indicated bootstrap support values ()
Zervas et al GBE
2900 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
of bchI before the crtB crtI genes in the first cluster For
Methylobacterium spp groups 2 and 3 show similar gene
syntenymdashthough there are a few differences in gene content
(table 2)mdashwhereas groups 1 and 4 appear to also share the
overall architecture of the PGCs For isolates WL9 and WL19
(both in group 1) the exact location of the crtB and crtI genes
is not known as they appear on a short contig by themselves
Annotating the assemblies on RAST revealed several house-
keeping genes present in the vicinity of these clusters which
further supported the notion of their chromosomal place-
ment The completed genomes resulting from the hybrid as-
semblies (strains WL1 WL3 WL4 and WL45) provided the
final evidence of the exact placement of the PGCs on
the chromosomes of all our isolates The different
architectures of the isolated PGCs are shown in figure 5
General descriptive statistics of the assembled complete
genomes are given in table 3 The hybrid assemblies resulted
in both complete genomes and plasmids Thus Rhizobium sp
WL3 contains two plasmids Roseomonas sp WL45 contains
two chromosomes and seven plasmids Alsobacter sp WL4
with two plasmids Methylobacterium sp WL1 with one
plasmid
Using Roseomonas sp WL45 as outgroup we generated
phylogenetic trees for the 16S acsF bchL bchY crtB pufL
and pufM genes as described previously (data not shown)
We calculated substitution rates from the root of the different
trees including the core proteome (CheckM) tree and com-
pared them to each other (table 4)
Discussion
A New Strategy to Generate Complete Genomes ofUnique Aerobic Anoxygenic Phototrophic Bacteria FromEnvironmental Isolates
Aiming to enhance our knowledge of the prevalence of aer-
obic anoxygenic phototrophic bacteria in the phyllosphere
we designed this study using wheat as the plant of choice
and employed a high-throughput multi-disciplinary approach
to identify isolate and analyze AAP strains From the initial
137 plates we ended up with 4480 bacterial colonies
gt500 of which were probable AAP strains After macroscopic
investigation of size shape color and texture of the colonies
we chose 129 strains for which we ran proteome profiling
using MALDI-TOF MS This approach has been shown to be
FIG 3mdashPairwise proteome comparisons (BLAST-matrix) of the 21 bacterial isolates of the analysis The gray-to-green color gradient indicates low to
high percentages of identity The bottom boxes show the presence of homologs within the same proteome (gray-to-red color gradient shows low to high
presence of homologs in the specific proteome)
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2901
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
able to reproducibly distinguish different taxa and in cases
where reference MALDI-TOF spectra are available in
the database be able to assign Genus andor Species names
(Madsen et al 2015) We then proceeded with whole ge-
nome sequencing on the Illumina Nextseq platform and for
the unique genomes we also proceeded with Oxford
Nanopore Technologies sequencing to generate complete
genomes and plasmids
Following this multi-disciplinary approach that relies on
classical microbiological techniques (plating morphology
etc) biochemical analyses (MALDI-TOF MS) and two differ-
ent types of high-throughput whole genome sequencing we
were able to generate draft and complete genomes of novel
AAP bacteria from the wheat phyllosphere in a relatively quick
(weeks) and efficient manner without losing novel biological
information in the process of reducing redundancy
Bacteriochlorophyll-Containing AAP Bacteria in the WheatPhyllosphere
By investigating the presence of bacteriochlorophylls in the
wheat phyllosphere we were able to identify 21 unique strains
harboring them the majority of which belonged to the genus
Methylobacterium Members of this genus are ubiquitous in
nature and have been detected in both soil and plant surfaces
(Green 2006) Targeting the bchY (chlorophyllide reductase
subunit Y) and pufM (reaction centre subunit M) genes in
phyllosphere samples of different plants it was shown that
the Methylobacteriaceae (alpha-Proteobacteria) comprised
28 of the total pufM sequences from the amplicon sequenc-
ing analysis (Atamna-Ismaeel and Finkel 2012a) Our results
are thus in accordance to these previous studies As
Methylobacteriaceae are indigenous inhabitants of the phyllo-
sphere and have consistently been shown to harbor PGCs it
may be that these photosystems are essential to the
Methylobacteria that inhabit this niche Interestingly in their
study Atamna-Ismaeel et al had negative results from their
bchY amplicon sequencing approach suggesting the absence
of bacteria that contain RC1 (bacteriochlorophyll-a containing
reaction centers) type photosystems Our results show that all
isolates contain the complete cassette of chlorophyllide syn-
thase subunit encoding genes (bchB bchC bchF bchG bchI
bchL bchM bchN bchP bchX bchY bchZ) as well as both the
pufL and pufM genes (table 2) Testing the same primers (Yutin
et al 2009) in silico against our bchY sequences we had pos-
itive results for all of them (data not shown)
Patterns of loss of the carotenoid genes have been ob-
served in other alpha and gamma proteobacteria like in
Rhodobacterales and Sphingomonadales (Zheng et al
2011) The absence of crtA results in the inability to perform
the final step in the spheroidene pathway where hydroxy-
spheroidene is converted to hydroxyspheroidenone
Moreover the absence of crtC in Methylobacterium spp of
groups 3 and 4 completely affects both the spheroidene and
the spirilloxanthin pathways as its product is essential for the
first step of both metabolic pathways starting from neuro-
sporene and lycopene respectively This means that strains
FIG 4mdashWhole genome average nucleotide identity (ANI) cladogram
showing percentages of k-mer identities between the isolates Taxa with
the same color and above 90 identity threshold are considered to belong
in the same species Beijerinckia indica ATCC-9039 was used as outgroup
of the analyses (not defined a priori)
Zervas et al GBE
2902 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Tab
le2
Gen
eC
onte
nt
of
the
Photo
synth
esis
Gen
eC
lust
ers
Iden
tified
inth
e21
Bac
terial
Isola
tes
Bact
eri
och
loro
ph
yll
Syn
thase
Gen
es
Caro
ten
oid
Bio
syn
thesi
sG
en
es
Lig
ht-
Harv
est
ing
Co
mp
lex
Gen
es
Str
ain
acs
Fb
chB
bch
Cb
chF
bch
Gb
chI
bch
Lb
chM
bch
Nb
chP
bch
Xb
chY
bch
Zcr
tAcr
tBcr
tCcr
tDcr
tEcr
tFcr
tIcr
tKp
ucC
pu
fAp
ufB
pu
fCp
ufL
pu
fMp
uh
A
Rh
izo
biu
msp
W
L3
Ro
seo
mo
nas
sp
WL4
5
Als
ob
act
er
sp
WL4
Gro
up
1M
eth
ylo
bact
eri
um
sp
WL9
Meth
ylo
bact
eri
um
sp
WL1
9
Meth
ylo
bact
eri
um
sp
WL6
9
Gro
up
2M
eth
ylo
bact
eri
um
sp
WL1
Meth
ylo
bact
eri
um
sp
WL2
Meth
ylo
bact
eri
um
sp
WL7
Meth
ylo
bact
eri
um
sp
WL1
8
Meth
ylo
bact
eri
um
sp
WL6
4
Gro
up
3M
eth
ylo
bact
eri
um
sp
WL6
Meth
ylo
bact
eri
um
sp
WL3
0
Meth
ylo
bact
eri
um
sp
WL9
3
Meth
ylo
bact
eri
um
sp
WL1
16
Meth
ylo
bact
eri
um
sp
WL1
19
Gro
up
4M
eth
ylo
bact
eri
um
sp
WL8
Meth
ylo
bact
eri
um
sp
WL1
2
Meth
ylo
bact
eri
um
sp
WL1
03
Meth
ylo
bact
eri
um
sp
WL1
20
Meth
ylo
bact
eri
um
sp
WL1
22
NO
TEmdash
Gra
yb
oxe
sin
dic
ate
pre
sen
ceo
fg
en
es
an
dw
hit
eb
oxe
sin
dic
ate
ab
sen
ceo
fg
en
es
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2903
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
of these two groups rely on the zeaxanthin pathway to pro-
duce carotenoid pigments and most likely nostoxanthin and
erythroxanthin which however do not participate in the
light-harvesting process but rather have a photoprotection
role (Noguchi et al 1992 Yurkov and Csotonyi 2009)
Phylogenetic Observations
In our study we sequenced and analyzed 21 aerobic anoxy-
genic photosynthetic bacteria most of which belonged to the
genus Methylobacterium the presence of which in the
environment has already been discussed previously
Methylobacterium spp are facultative methylotrophs and
can utilize a variety of C1 C2 C3 and C4 compounds
Such compounds are often found in the phyllosphere due
to plant metabolism (Vorholt 2012 Remus-Emsermann
et al 2014) Methylobacterium spp show a wide range in
chromosome size and number of plasmids they contain
(Marx et al 2012) The genus is represented well in
GenBank with 100 complete or draft genomes (last accessed
April 2019) However most of the sequences do not have
FIG 5mdashArchitecture of the different photosynthetic gene clusters Arrows indicate the orientation of the genes Sizes are not representatives of gene
length Only the genes shared by all PGCs were shown Genes that may be missing from some strains are denoted with parentheses on the figurersquos legend
Table 3
General Descriptive Statistics of the Four Complete AAP Bacterial Genomes
Taxon Genome Size (bp) GC Number of Genes Plasmids Plasmids Size (bp) GC
Rhizobium sp WL3 4568855 6160 4450 1 700631 586
mdash mdash mdash 2 85350 575
Roseomonas sp WL45 4533887 708 5042 1 262815 655
1011854 707 1138 2 240556 662
mdash mdash mdash 3 152922 658
mdash mdash mdash 4 85279 646
mdash mdash mdash 5 84774 661
mdash mdash mdash 6 83273 656
mdash mdash mdash 7 65860 648
Alsobacter sp WL4 5405668 679 5013 1 27178 620
mdash mdash mdash 2 9451 606
Methylobacterium sp WL1 6215463 691 6431 1 35731 638
Zervas et al GBE
2904 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
species affiliation (fig 4) Thus we were not able to assign any
of the 18 isolates to specific Methylobacterium species apart
perhaps from group 2 (WL1 WL2 WL7 WL18 WL64) which
is placed as sister to M brachiatum in the MASH-ANI analysis
Even in this case though M pseudosasicola BL36 is also
placed in the same clade as M brachiatum which raises the
question of proper nomenclature This is further evidenced in
the same tree (cyanide clade below) where seven species
share gt90 of their genome For the remaining isolates
WL19 is closely affiliated to other Methylobacterium strains
that have not been identified A similar picture is shown for
WL69 On the other hand WL19 appears to be a novel spe-
cies with unique genomic features and so do the strains from
groups 3 and 4 (red clade) Thus despite how ubiquitous
Methylobacterium is in nature it is clear that part of its diver-
gence is still missing Moreover a critical evaluation of the
current nomenclature of the genus would be beneficial based
on the whole genome sequences because closely related
genomes (MASH-ANI identitygt 97) have been given differ-
ent species affiliations
Apart from this genus we also identified a Rhizobium sp
a Roseomonas sp and a novel Alsobacter strain (WL4) that
belongs in the Methylocystaceae a family of methanotroph
bacteria This is the first time that AAP bacteria from this
family have been isolated and their complete genomes assem-
bled Its closely related Alsobacter spp have been shown to
exhibit high Thallium tolerance (Bao et al 2006) and it will be
interesting to identify heavy-metal resistance genes in non-
heavy-metal polluted phyllosphere samples It is worth men-
tioning that if WL4 contains heavy-metal resistance genes on
top of the PGC this will have been shown for the first time
and its potential to bioremediate heavy-metal contaminated
areas if applied as plant growth promotion and environmental
remediation agent is of high value and needs to further be
explored
Molecular Evolution of the PGCs
For all gene trees of the PGC we observed that Rhizobium sp
WL3 evolves faster than the other isolates followed by
Table 4
Substitutions Per Site in the Different Phylogenetic Analyses
Strain 16S acsF bchL bchY crtB pufL pufMsuper-matrix
CheckM
Roseomonas sp WL45 0096 0147 0100 0235 0250 0095 0132 0160 0184
Rhizobium sp WL3 0184 0622 0767 0690 1370 0343 0525 0580 0372
Alsobacter sp WL4 0151 0462 0233 0490 0870 0258 0334 0380 0403
Methylobacterium sp WL9 0226 0418 0372 0420 0670 0270 0319 0364 0477
Methylobacterium sp WL19 0221 0413 0300 0460 0600 0233 0346 0364 0468
Methylobacterium sp WL69 0231 0373 0333 0470 0620 0260 0358 0368 0468
Methylobacterium sp WL1 0233 0422 0344 0485 0480 0280 0385 0380 0459
Methylobacterium sp WL2 0233 0422 0344 0485 0480 0280 0385 0380 0459
Methylobacterium sp WL7 0238 0427 0344 0480 0480 0278 0385 0380 0459
Methylobacterium sp WL18 0238 0427 0344 0480 0480 0280 0385 0380 0459
Methylobacterium sp WL64 0233 0422 0344 0480 0485 0280 0385 0380 0459
Methylobacterium sp WL6 0199 0378 0256 0425 0680 0208 0311 0340 0455
Methylobacterium sp WL30 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL93 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL116 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL119 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL8 0188 0373 0258 0430 0680 0230 0358 0348 0459
Methylobacterium sp WL12 0188 0373 0256 0430 0680 0230 0358 0348 0459
Methylobacterium sp WL103 0188 0373 0256 0430 0850 0230 0358 0364 0459
Methylobacterium sp WL120 0188 0373 0258 0430 0680 0233 0358 0348 0459
Methylobacterium sp WL122 0188 0373 0256 0430 0680 0230 0358 0348 0459
Susbstitutions per site
Gro
up1
Gro
up 2
Gro
up 3
Gro
up 4
NOTEmdashAll values correspond to distances from the root where Roseomonas sp WL45 was chosen as outgroup for all the alignments
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2905
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Alsobacter sp WL4 Both these isolates also show much
higher substitution rates compared with their respective 16S
genes and core proteomes (checkM) which indicates that the
PGCs are under relaxed selection pressure In all instances the
phylogenetic trees constructed showed the same topology
and the same grouping of the Methylobacterium isolates
which suggests that the PGCs were incorporated in the
genomes of these strains relatively early in their evolution If
the PGC was free to move laterally then we would have ob-
served very different phylogenetic trees for its genes com-
pared with the 16S and core proteome analyses
Interestingly these observations do not reflect on the archi-
tecture of the PGC where Roseomonas sp WL45 and
Rhizobium sp WL3 contain the PGC in one continuous re-
gion whereas Alsobacter sp WL4 and Methylobacterium sp
WL1 have it in two pieces
For the 18 Methylobacterium isolates group 1 and 2
strains (WL9 WL19 WL69 and WL1 WL2 WL7 WL18
WL64) evolve consistently at different rates compared
with the groups 3 and 4 It appears that there is relatively
strong selection pressure on bchL pufL and pufM which
in all cases are more similar to each other and also show
slower molecular evolution rates compared with ascF
bchY and crtB which in all Methylobacterium isolates
evolve at least twice as fast compared with the 16S
gene (table 4) They also show fewer substitutions per
site compared with the core proteome analysis This result
partially contrasted with previously adopted methods of
using bchY as a marker gene for RC1 clusters (eg
Atamna-Ismaeel and Finkel 2012a) On the other hand
the use of pufM appears more sensible It would how-
ever be more suitable to investigate the possibility of us-
ing other genes for metagenome amplicon sequencing
approaches such as pufL or bchL which show a smaller
degree of divergence in different genera thus universal
primers might be more successful in detecting a wide va-
riety of AAP bacteria in environmental samples
Interestingly while Groups 1 and 2 isolates exhibit higher
substitution rates of the PGC genes compared with
Groups 3 and 4 this is not the case for crtB which evolves
a lot slower and also pufM which shows similar molec-
ular evolution in the genus Overall it seems that the
PGCs were incorporated in ancestral strains that later di-
verged to give rise to the different AAP taxa This is fur-
ther corroborated by the fact that almost all phylogenetic
trees (eg fig 5) are consistent and in agreement with
both the 16S phylogenetic tree (fig 1) as well as the phy-
logenomic tree (fig 2) If the genes of the PGCs had been
transferred to these AAP several times there would have
been significant differences both in the sequence level
and the phylogenetic analysis of the different genes
which would be expected to result in different topologies
and diverse distances from the root even for closelymdash
based on 16S analysismdashrelated strains
The Significance of AAP Bacteria in Wheat Phyllosphere
In this study we isolated for the first time a variety of AAP
bacteria that are present in the wheat phyllosphere adding to
the metagenomics knowledge provided earlier from five other
plants (Atamna-Ismaeel and Finkel 2012a 2012b) The phyllo-
sphere is a rather harsh environment with little water avail-
ability scarce resources very few suitable locations extreme
temperature difference between day and night and high
dosages of UV radiation (Vorholt 2012) AAP bacteria have
the ability to utilize light in photophosphorylation processes
through which they produce ATP and at the same time funnel
the few available nutrients in other metabolic pathways
(Yurkov and Hughes 2017) This has been shown to give
them an edge in vitro after exposure to light compared with
nonAAP bacteria (Ferrera et al 2011) Considering that phyl-
losphere bacteria protect the plant from pathogenic microbes
by secreting metabolites and by occupying the available
niches being able to persevere and outgrow other bacteria
gives a significant edge to AAP taxa and probably points
toward the suggestion of positive selection of such bacteria
in their phyllosphere by the plants themselves as they appear
to be absent from the soil (Atamna-Ismaeel and Finkel
2012b) At this point it is not clear exactly how AAP bacteria
benefit the plant especially since the phyllosphere is a com-
plex habitat with an ever-changing harsh environment In the
past years great efforts have been carried out on how to
improve crop production by studying the rhizosphere It has
also been suggested that the phyllosphere communities play a
significant role in plant health and growth which ultimately
leads to increased yields (Lindow and Brandl 2003) Given the
current projections of world population growth and the sup-
ply of food (UN 2017) it becomes evident that more in-depth
studies investigating the phyllosphere of plants and especially
crops and identifying the underlying drivers of the bacterial
communities and their interactions with their plant hosts is of
paramount importance Thus by isolating maintaining and
analyzing the genomes of phyllosphere isolates we may iden-
tify strains that may prove useful as plant growth promoting
agents in future field trials investigating the effect of phyllo-
sphere inoculation on plant growth health and yield
Supplementary Material
Supplementary data are available at Genome Biology and
Evolution online
Acknowledgments
The authors would like to thank Tue Sparholt Jorgensen for
collecting the wheat sample The authors also want to thank
Tina Thane (AU) Tanja Begovic (AU) and Margit Frederiksen
(NRCWE) for capable laboratory assistance We thank Michal
Koblızek for providing the initial model of the colony infrared
imaging system This work was funded by a Villum
Zervas et al GBE
2906 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Experiment grant (no 17601) and a Marie Skłodowska-Curie
AIAS-COFUND fellowship (EU-FP7 program Grant
Agreement No 609033) awarded to YZ
Author Contributions
AZ YZ LHH designed the study YZ isolated the strains
performed colony infrared imaging and extracted DNA for
Illumina sequencing YZ and AMM performed the MALDI-
TOF MS analysis AZ extracted DNA and performed ONT
Sequencing AZ carried out the bioinformatics and phyloge-
netic analyses AZ and YZ drafted the manuscript AZ
YZ AMM and LHH revised the manuscript for
intellectual content All authors read and approved the
manuscript
Literature Cited
Atamna-Ismaeel N Finkel O 2012a Bacterial anoxygenic photosynthesis
on plant leaf surfaces Environ Microbiol Rep 4(2)209ndash216
Atamna-Ismaeel N Finkel OM 2012b Microbial rhodopsins on leaf sur-
faces of terrestrial plants Environ Microbiol 14(1)140ndash146
Aziz RK et al 2008 The RAST Server rapid annotations using subsystems
technology BMC Genomics 9(1)75
Bankevich A et al 2012 SPAdes a new genome assembly algorithm and
its applications to single-cell sequencing J Comput Biol
19(5)455ndash477
Bao Z Sato Y Kubota M Ohta H 2006 Isolation and characterization of
thallium-tolerant bacteria from heavy metal-polluted river sediment
and non-polluted soils Microb Environ 21(4)251ndash260
Beatie GA 2002 Leaf surface waxes and the process of leaf colonization
by microorganisms In Lindow SE Hecht-Poinar EI Elliott VJ editors
Phyllosphere Microbiology St Paul (USA) APS Press pp 3ndash26
Beattie GA Lindow SE 1999 Bacterial colonization of leaves a spectrum
of strategies Phytopathology 89(5)353ndash359
Brettin T et al 2015 RASTtk a modular and extensible implementation of
the RAST algorithm for building custom annotation pipelines and an-
notating batches of genomes Sci Rep 58365
Carvalho SD Castillo JA 2018 Influence of light on plantndashphyllosphere
interaction Front Plant Sci 91482
Ferrera I Gasol JM Sebastian M Hojerova E Koblızek M 2011
Comparison of growth rates of aerobic anoxygenic phototrophic bac-
teria and other bacterioplankton groups in coastal Mediterranean wa-
ters Appl Environ Microbiol 77(21)7451ndash7458
Ferrera I Sanchez O Kolarova E Koblızek M Gasol JM 2017 Light
enhances the growth rates of natural populations of aerobic anoxy-
genic phototrophic bacteria ISME J 11(10)2391ndash2393
Gorlach K Shingaki R Morisaki H Hattori T 1994 Construction of eco-
collection of paddy field soil bacteria for population analysis J Gen
Appl Microbiol 40(6)509ndash517
Gourion B Rossignol M Vorholt JA Lindow SE 2006 A proteomic study
of Methylobacterium extorquens reveals a response regulator essential
for epiphytic growth wwwintegratedgenomicscom Accessed
November 26 2018
Green PN 2006 Methylobacterium In Dworkin M Falkow S Rosenberg
E Schleifer K Stackebrandt E editors The prokaryotes ndash a handbook
on the biology of bacteria New York Springer p 257ndash265
Hauruseu D Koblızek M 2012 Influence of light on carbon utilization in
aerobic anoxygenic phototrophs Appl Environ Microbiol
78(20)7414ndash7419
Katoh K Misawa K Kuma K Miyata T 2002 MAFFT a novel method for
rapid multiple sequence alignment based on fast Fourier transform
Nucleic Acids Res 30(14)3059ndash3066
Knief C et al 2012 Metaproteogenomic analysis of microbial communi-
ties in the phyllosphere and rhizosphere of rice ISME J
6(7)1378ndash1390
Knief C Frances L Vorholt JA 2010 Competitiveness of diverse methyl-
obacterium strains in the phyllosphere of arabidopsis thaliana and
identification of representative models including M extorquens
PA1 Microb Ecol 60(2)440ndash452
Kwak MJ et al 2014 Genome information of Methylobacterium oryzae a
plantndashprobiotic methylotroph in the phyllosphere PLoS One
9(9)e106704
Lindow SE Brandl MT 2003 Microbiology of the phyllosphere Appl
Environ Microbiol 69(4)1875ndash1883
Madsen AM Zervas A Tendal K Nielsen JL 2015 Microbial diversity in
bioaerosol samples causing ODTS compared to reference bioaerosol
samples as measured using Illumina sequencing and MALDI-TOF
Environ Res 140255ndash267
Martin M 2011 Cutadapt removes adapter sequences from high-
throughput sequencing reads EMBnet J 17(1)10ndash12
Marx CJ et al 2012 Complete genome sequences of six strains of the
genus Methylobacterium b J Bacteriol 194(17)4746
Noguchi T Hayashi H Shimada K Takaichi S Tasumi M 1992 In vivo
states and functions of carotenoids in an aerobic photosynthetic bac-
terium Erythrobacter longus Photosynth Res 31(1)21ndash30
Olm MR Brown CT Brooks B Banfield JF 2017 dRep a tool for fast and
accurate genomic comparisons that enables improved genome recov-
ery from metagenomes through de-replication ISME J
11(12)2864ndash2868
Overbeek R et al 2014 The SEED and the Rapid Annotation of microbial
genomes using Subsystems Technology (RAST) Nucl Acids Res
42(D1)D206ndashD214
Parks DH Imelfort M Skennerton CT Hugenholtz P Tyson GW 2015
CheckM assessing the quality of microbial genomes recovered from
isolates single cells and metagenomes Genome Res 25(7)1043ndash1055
Remus-Emsermann MNP et al 2014 Spatial distribution analyses of nat-
ural phyllosphere-colonizing bacteria on Arabidopsis thaliana revealed
by fluorescence in situ hybridization Environ Microbiol
16(7)2329ndash2340
Stamatakis A 2006 RAxML-VI-HPC maximum likelihood-based phyloge-
netic analyses with thousands of taxa and mixed models
Bioinformatics 22(21)2688ndash2690
UN 2017 World population prospects the 2017 revision Data Booklet
httpspopulationunorgwppPublicationsFilesWPP2019_DataBook-
letpdf
Vesth T Lagesen K Acar euroO Ussery D 2013 CMG-biotools a free work-
bench for basic comparative microbial genomics PLoS One
8(4)e60120
Vorholt JA 2012 Microbial life in the phyllosphere Nat Rev Microbiol
10(12)828ndash840
Wick RR Judd LM Gorrie CL Holt KE 2017 Unicycler resolving bacterial
genome assemblies from short and long sequencing reads PLoS
Comput Biol 13(6)e1005595
Wilson M Lindow SE 1994a Coexistence among epiphytic bacterial pop-
ulations mediated through nutritional resource partitioning Appl
Environ Microbiol 604468ndash77
Wilson M Lindow SE 1994b Ecological similarity and coexistence of epi-
phytic ice-nucleating (ice) pseudomonas syringae strains and a non-
ice-nucleating (ice) biological control agent Appl Environ Microbiol
60(9)3128ndash3137
Woodward FI Lomas MR 2004 Vegetation dynamics ndash simulating
responses to climatic change Biol Rev 79(3)643ndash670
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2907
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Xu J Gordon JI 2003 Honor thy symbionts Proc Natl Acad Sci USA
100(18)10452ndash10459
Yurkov V Csotonyi JT 2009 New light on aerobic anoxygenic photo-
trophs In Hunter NC Daldal F Thurnauer MC Beatty JT editors
The purple photosynthetic bacteria Dordrecht Springer thorn Business
Media BV p 31ndash55 doi101007978-1-4020-8815-5_3
Yurkov V Hughes E 2017 Aerobic Anoxygenic Phototrophs four
decades of mystery In Hallenbeck PC editor Modern topics in the
phototrophic prokaryotes environmental and applied aspects p
193ndash214
Yutin N et al 2009 BchY-based degenerate primers target all types of
anoxygenic photosynthetic bacteria in a single PCR Appl Environ
Microbiol 75(23)7556ndash7559
Zeng Y Feng F Medova H Dean J Koblizek M 2014 Functional
type 2 photosynthetic reaction centers found in the rare bacterial phy-
lum Gemmatimonadetes Proc Natl Acad Sci USA 111(21)7795ndash7800
Zheng Q et al 2011 Diverse arrangement of photosynthetic gene
clusters in aerobic anoxygenic phototrophic bacteria PLoS One
6(9)e25050
Associate editor Esperanza Martinez-Romero
Zervas et al GBE
2908 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
- evz204-TF1
- evz204-TF2
-
capabilities of the isolated strains was performed using the
same setup Strains that passed the verification step were
further restreaked onto 15 strength R2A plates to ensure
pure colonies of single AAP strains
Selection of AAP Strains and Whole Genome Sequencing
The isolated AAP strains were grouped and dereplicated using
a MALDI-TOF MS (Microflex LT Bruker Daltonics Bremen
Germany) before performing whole genome sequencing to
reduce the cost while maintaining the biodiversity Briefly a
toothpick was used to transfer a small amount of an AAP
bacterial colony onto the target plate (MSP 96 polished steel
Bruker) that was evenly spread out and formed a thin layer of
biomass on the steel plate The sample was then overlaid with
70 formic acid and allowed for air dry before addition of
1lL MALDI-MS matrix solution (a-cyano-4-hydroxycinnamic
acid SigmandashAldrich) A bacterial standard with well-
characterized peaks (Bruker Daltonics) was used to calibrate
the instrument The standard method ldquoMBT_AutoXrdquo was ap-
plied to obtain proteome profiles within the mass range of 2ndash
20 kDa using the flexControl software (Bruker) The
flexAnalysis software (Bruker) was used to smooth the data
subtract baseline and generate main spectra (MSP) followed
by a hierarchical clustering analysis with the MALDI Biotyper
Compass Explorer software which produced a dendrogram
as the output for visual inspection of similarities between
samples An empirical distance value of 50 was used as the
cutoff for defining different groups on the dendrogram cor-
responding to different strainsspecies
From each group on the dendrogram one isolate was cho-
sen and restreaked on 1=2 R2A plates and incubated at room
temperature for one week Prior to DNA isolation the plates
were again tested for the presence of light absorbing pigments
as previously described Bacterial colonies were scraped from
the plates using a loop and immersed in 400mL PBS solution
vortexed until the cells had separated spun down in a table-top
centrifuge at 10000 g at 4 C the supernatant was removed
and the pellets were redissolved in milliQ H2O High molecular
weight DNA was extracted from each strain using a modified
Masterpure Complete DNA amp RNA Purification Kit (Lucigen) by
replacing the elution buffer with a 10mM TrisndashHClmdash50mM
NaCl (pH 75ndash80) solution at the final step DNA concentration
was quantified on a Qubit 20 (Life Technologies) using the
Broad Range DNA Assay kit and DNA quality was measured
on a Nanodrop 2000C (Thermo Scientific) Whole genome
shotgun sequencing was performed on all AAP strains on the
Illumina Nextseq platform in house using the 2150bp chem-
istry Selected strains were also sequenced on the MinION plat-
form (Oxford Nanopore Technologies UK) in house using the
Rapid Barcoding Kit (SBQ-RBK004) and the FLO-MIN106 flow
cell (R941) following the manufacturerrsquos instructional
manuals
Genome Assembly and Analyses
The Illumina pair end reads were trimmed for quality and
ambiguities using Cutadapt (Martin 2011) and assembled
using SPAdes (Bankevich et al 2012) The resulting assem-
blies were analyzed using dRep (Olm et al 2017) to compare
their average nucleotide identities (ANI) in order to identify
clusters of closely related taxa The assemblies were
imported in Geneious R1125 (Biomatters Ltd) and PGCs
were annotated using the Roseobacter litoralis strain
Och149 plasmid pRL0149 (CP002624) as reference under
relaxed similarity criteria (40) The suggested genes of
probable PGCs were analyzed using BLAST (megablast
BlastN) to verify their annotations For full genome annota-
tions the assemblies were uploaded on RAST (Aziz et al
2008 Overbeek et al 2014 Brettin et al 2015) The pre-
dicted translated protein coding genes were imported in
CMG-Biotools (Vesth et al 2013) for pairwise proteome
comparisons using the native blastmatrix program
Hybrid assemblies of pair-end Illumina reads and long
Oxford Nanopore Technologies reads were performed using
the Unicycler assembler (Wick et al 2017) utilizing the bold
assembly mode The resulting complete genomes were
imported in Geneious where their circularity was confirmed
by mapping-to-reference runs using the short and long reads
from the respective hybrid assemblies and the native
Geneious mapper under default settings in the Mediumndash
Low sensitivity mode
Phylogenetic Analyses
Identification of the 16S ribosomal DNA sequences was per-
formed by rnammer in CMG-Biotools Data sets containing
the different genes from the identified PGCs were created in
Geneious The resulting sequences were aligned using MAFFT
v7388 (Katoh et al 2002) and it FFT-NS-I x1000 algorithm
with default settings as implemented in Geneious
Phylogenetic trees were created using RAxML (Stamatakis
2006) employing the GTR-GAMMA nucleotide substitution
model and the rapid bootstrapping and search for best scor-
ing ML-tree with 100 bootstraps replicates and starting from
a random tree Analysis of the PGCs was performed on five
selected genes that were present in all isolates namely acsF
bchL bchY crtB pufL and pufM Individual gene alignments
and phylogenetic trees were performed as described above
A super-matrix (8240 characters) comprised of the total
seven gene alignments was created using the built-in
ldquoConcatenate alignmentsrdquo function in Geneious and a phy-
logenetic tree was created as above All trees were visually
inspected for disagreements Finally a phylogenomic tree
was constructed in RAxML employing the GAMMA
BLOSUM62 substitution model and the rapid bootstrapping
and search for best scoring ML-tree algorithm with 100 boot-
straps replicates and starting from a random tree using as
input the core protein alignment consisting of 6988
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2897
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
characters created with CheckM (Parks et al 2015) and its
lineage_wf algorithm with default settings Substitution rates
were calculated as distances from the root using the legend
provided by the phylogenetic trees
Results
Identification and Isolation of AAP Bacteria from thePhyllosphere
The 129 phyllosphere isolates that were tested positive for the
presence of bacteriochlorophyll-a were placed in 21 groups
according to their protein mass profiles based on MALDI-TOF
MS (supplementary fig MALDI-TOF Supplementary Material
online) A representative from each group was sequenced
(Illumina) and their genomes assembled (SPAdes) WL1 and
WL2 were from the same MALDI-TOF group and their
genomes are identical (ANI 100) confirming the validity
of MALDI-TOF MS for grouping similar isolates Information
about their phylogeny total genome size and genome as-
sembly statistics are presented in table 1
Analysis of the 16S genes showed that 17 isolates
belonged in Methylobacterium (separated in four distinct
groups) whereas the remaining three belonged to
Rhizobium Roseomonas and one novel species was affiliated
to the Methylocystaceae family (WL4) with most hits belong-
ing to uncultured environmental strains The only character-
ized hits of WL4 belonged to Methylosinus trichosporium
(Gorlach et al 1994) and Alsobacter metallidurans a recently
characterized novel genus novel species (Bao et al 2006)
However in both instances the similarity was low
(985) We included strains of both species and created
a 16S phylogenetic tree (fig 1) which placed Alsobacter sp
WL4 as sister to the Alsobacter-Methylosinus clade
The core proteome analysis (fig 2mdashCheckM) including M
trichosporium OB3b and Alsobacter sp SH9 established the
phylogeny of strain WL4 and it was highly supported as sister
to Alsobacter sp SH9 (BSfrac14 100) both of which were
distinct from M trichosporium OB3b (BSfrac14 100)
Methylobacterium strains were still placed into four distinct
groups (Group 1 WL9 WL19 WL69 Group 2 WL1 WL2
WL7 WL18 WL64 Group 3 WL6 WL30 WL93 WL116
WL119 Group 4 WL8 WL12 WL103 WL120) This was
further supported by the whole proteome comparison
(fig 3mdashBLAST matrix) For Groups 3 and 4 the average
core protein content is 4500 proteins whereas for Group
2 it is close to 4000 This may signify the presence of
either more or larger plasmids that characterize the strains
of this group For Group 1 this number falls even more as
in this case there clearly are three distinct species that make
up the group
We expanded the analysis to include 100 complete and
draft Methylobacterium spp genomes available in GenBank
(last assessed April 2019) and also other representatives from
the Methylocystaceae aiming at further establishing strain
WL4 in the genus Alsobacter and also providing greater res-
olution on the Methylobacterium isolates with a special focus
on the divergent group 1 Information about the included
genomes is given in supplementary table 1 Supplementary
Material online We performed whole genome comparisons
using ANI (fig 4mdashdRep)
The analysis showed that Group 1 isolates belonged in
different Methylobacterium species groups 3 and 4 were dif-
ferent strains of the same species (MASH ANI identity be-
tween groupsgt 90) whereas group 3 isolates are closely
related strains possibly belonging to Methylobacterium bra-
chiatum Strain WL4 is again placed close to the two identical
Alsobacter genomes (Alsobacter sp SH9 and A metallidurans
PVZS01) far away from the three remaining species from
within the Methylocystaceae (M trichosporium OB3b
Methylocystis rosea GW6 Methylocystis bryophila S285)
However the MASH-ANI identity between WL4 and
Alsobacter is below the 80 threshold of genus assignment
Therefore we took the top 200 megablast hits for WL4rsquos 16S
region (information provided in supplementary table 2
Supplementary Material online) aligned them including strain
WL4 as mentioned previously and created a phylogenetic tree
of the 201 sequences (supplementary fig 1 Supplementary
Material online) The results clearly show that WL4 belongs in
Alsobacter whereas Methylosinus spp form their own dis-
tinct cluster
Table 1
Genome Statistics and Phylogenetic Information of the 21 Isolates that
were Illumina Sequenced in This Study
Strain Taxonomy Contigs Genome Size N50 GC
WL1 Methylobacterium 654 6258106 19384 6892
WL2 Methylobacterium 298 6292310 52057 6899
WL3 Rhizobium 55 5333961 210875 6114
WL4 Alsobacter 139 5412272 86934 6792
WL6 Methylobacterium 708 5549874 15034 6972
WL7 Methylobacterium 431 6059272 36478 6904
WL8 Methylobacterium 225 5322599 56100 6969
WL9 Methylobacterium 221 4670253 42885 6741
WL12 Methylobacterium 271 5545758 53038 694
WL18 Methylobacterium 1373 5796709 7880 6864
WL19 Methylobacterium 330 5177712 31633 6698
WL30 Methylobacterium 377 5636306 38077 6978
WL45 Roseomonas 2056 5961800 5511 6954
WL64 Methylobacterium 365 6908402 48840 6834
WL69 Methylobacterium 225 4693485 43639 6951
WL93 Methylobacterium 719 5642324 16902 6945
WL103 Methylobacterium 1327 5167285 7154 6897
WL116 Methylobacterium 1173 5445855 8413 6943
WL119 Methylobacterium 263 5709790 45485 6958
WL120 Methylobacterium 395 5266148 26698 6954
WL122 Methylobacterium 2024 4508072 3666 6877
Zervas et al GBE
2898 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Molecular Evolution of the PGC
The PGCs identified in the sequenced isolates all belong to
RC-II type meaning they contain bacteriochlorophyll-a and
possess pheophytinmdashquinone type reaction centers All
strains possess a full bacteriochlorophyll synthase gene set
except for Alsobacter sp WL4 that is lacking the bchM
gene (table 2)
For the carotenoid biosynthesis genes all strains possess
crtB whereas crtA is only present in Alsobacter sp WL4 and
Rhizobium sp strain WL3 For the crtC and crtK genes an
interesting pattern is observed Methylobacterium group 1
and group 2 strains contain crtC (with three exceptions dis-
cussed below) whereas group 3 and 4 strains completely lack
the gene On the other hand crtK is only present in
FIG 1mdashPhylogenetic tree of the 16S genes of the sequenced isolates Methylosinus trichosporium and Alsobacter metallidurans were included to better
pinpoint the phylogenetic placement of strain WL4 The legend shows substitution rates from the root of the tree and the node labels indicate bootstrap
support values ()
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2899
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Methylobacterium group 3 strains and missing from groups 1
2 and 4 Similar patterns are evidenced in pufB which is ab-
sent from Methylobacterium groups 3 and 4 Rhizobium sp
WL3 and Roseomonas sp WL45 and pufC missing from
Methylobacterium group 4 as well as the novel strain WL4
The remaining genes were found ubiquitously in all strains of
the analysis with only a few exceptions (table 2)
The PGCs were consistently found on the longer contigs of
the initial assemblies of Illumina data produced by SPAdes
(for the strains with fewer than 500 contigs) (table 1) In
Rhizobium sp WL3 and Roseomonas sp WL45 the photosyn-
thetic genes form one continuous cluster with different
however architectures In Alsobacter sp WL4 the genes are
organized in two clusters similar to the Methylobacterium
spp isolates In these cases the genes were found in the
middle of the contigs ruling out the possibility of PGC being
organized into two clusters as an artifact of the assembly
Unique to the architecture present in WL4 is the presence
FIG 2mdashPhylogenomic tree based on the core proteome (CheckM) of the 21 pure bacterial isolates of the analysis including Methylosinus trichosporium
OB3B and Alsobacter sp SH9 The legend shows substitution rates from the root of the tree and the node labels indicated bootstrap support values ()
Zervas et al GBE
2900 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
of bchI before the crtB crtI genes in the first cluster For
Methylobacterium spp groups 2 and 3 show similar gene
syntenymdashthough there are a few differences in gene content
(table 2)mdashwhereas groups 1 and 4 appear to also share the
overall architecture of the PGCs For isolates WL9 and WL19
(both in group 1) the exact location of the crtB and crtI genes
is not known as they appear on a short contig by themselves
Annotating the assemblies on RAST revealed several house-
keeping genes present in the vicinity of these clusters which
further supported the notion of their chromosomal place-
ment The completed genomes resulting from the hybrid as-
semblies (strains WL1 WL3 WL4 and WL45) provided the
final evidence of the exact placement of the PGCs on
the chromosomes of all our isolates The different
architectures of the isolated PGCs are shown in figure 5
General descriptive statistics of the assembled complete
genomes are given in table 3 The hybrid assemblies resulted
in both complete genomes and plasmids Thus Rhizobium sp
WL3 contains two plasmids Roseomonas sp WL45 contains
two chromosomes and seven plasmids Alsobacter sp WL4
with two plasmids Methylobacterium sp WL1 with one
plasmid
Using Roseomonas sp WL45 as outgroup we generated
phylogenetic trees for the 16S acsF bchL bchY crtB pufL
and pufM genes as described previously (data not shown)
We calculated substitution rates from the root of the different
trees including the core proteome (CheckM) tree and com-
pared them to each other (table 4)
Discussion
A New Strategy to Generate Complete Genomes ofUnique Aerobic Anoxygenic Phototrophic Bacteria FromEnvironmental Isolates
Aiming to enhance our knowledge of the prevalence of aer-
obic anoxygenic phototrophic bacteria in the phyllosphere
we designed this study using wheat as the plant of choice
and employed a high-throughput multi-disciplinary approach
to identify isolate and analyze AAP strains From the initial
137 plates we ended up with 4480 bacterial colonies
gt500 of which were probable AAP strains After macroscopic
investigation of size shape color and texture of the colonies
we chose 129 strains for which we ran proteome profiling
using MALDI-TOF MS This approach has been shown to be
FIG 3mdashPairwise proteome comparisons (BLAST-matrix) of the 21 bacterial isolates of the analysis The gray-to-green color gradient indicates low to
high percentages of identity The bottom boxes show the presence of homologs within the same proteome (gray-to-red color gradient shows low to high
presence of homologs in the specific proteome)
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2901
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
able to reproducibly distinguish different taxa and in cases
where reference MALDI-TOF spectra are available in
the database be able to assign Genus andor Species names
(Madsen et al 2015) We then proceeded with whole ge-
nome sequencing on the Illumina Nextseq platform and for
the unique genomes we also proceeded with Oxford
Nanopore Technologies sequencing to generate complete
genomes and plasmids
Following this multi-disciplinary approach that relies on
classical microbiological techniques (plating morphology
etc) biochemical analyses (MALDI-TOF MS) and two differ-
ent types of high-throughput whole genome sequencing we
were able to generate draft and complete genomes of novel
AAP bacteria from the wheat phyllosphere in a relatively quick
(weeks) and efficient manner without losing novel biological
information in the process of reducing redundancy
Bacteriochlorophyll-Containing AAP Bacteria in the WheatPhyllosphere
By investigating the presence of bacteriochlorophylls in the
wheat phyllosphere we were able to identify 21 unique strains
harboring them the majority of which belonged to the genus
Methylobacterium Members of this genus are ubiquitous in
nature and have been detected in both soil and plant surfaces
(Green 2006) Targeting the bchY (chlorophyllide reductase
subunit Y) and pufM (reaction centre subunit M) genes in
phyllosphere samples of different plants it was shown that
the Methylobacteriaceae (alpha-Proteobacteria) comprised
28 of the total pufM sequences from the amplicon sequenc-
ing analysis (Atamna-Ismaeel and Finkel 2012a) Our results
are thus in accordance to these previous studies As
Methylobacteriaceae are indigenous inhabitants of the phyllo-
sphere and have consistently been shown to harbor PGCs it
may be that these photosystems are essential to the
Methylobacteria that inhabit this niche Interestingly in their
study Atamna-Ismaeel et al had negative results from their
bchY amplicon sequencing approach suggesting the absence
of bacteria that contain RC1 (bacteriochlorophyll-a containing
reaction centers) type photosystems Our results show that all
isolates contain the complete cassette of chlorophyllide syn-
thase subunit encoding genes (bchB bchC bchF bchG bchI
bchL bchM bchN bchP bchX bchY bchZ) as well as both the
pufL and pufM genes (table 2) Testing the same primers (Yutin
et al 2009) in silico against our bchY sequences we had pos-
itive results for all of them (data not shown)
Patterns of loss of the carotenoid genes have been ob-
served in other alpha and gamma proteobacteria like in
Rhodobacterales and Sphingomonadales (Zheng et al
2011) The absence of crtA results in the inability to perform
the final step in the spheroidene pathway where hydroxy-
spheroidene is converted to hydroxyspheroidenone
Moreover the absence of crtC in Methylobacterium spp of
groups 3 and 4 completely affects both the spheroidene and
the spirilloxanthin pathways as its product is essential for the
first step of both metabolic pathways starting from neuro-
sporene and lycopene respectively This means that strains
FIG 4mdashWhole genome average nucleotide identity (ANI) cladogram
showing percentages of k-mer identities between the isolates Taxa with
the same color and above 90 identity threshold are considered to belong
in the same species Beijerinckia indica ATCC-9039 was used as outgroup
of the analyses (not defined a priori)
Zervas et al GBE
2902 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Tab
le2
Gen
eC
onte
nt
of
the
Photo
synth
esis
Gen
eC
lust
ers
Iden
tified
inth
e21
Bac
terial
Isola
tes
Bact
eri
och
loro
ph
yll
Syn
thase
Gen
es
Caro
ten
oid
Bio
syn
thesi
sG
en
es
Lig
ht-
Harv
est
ing
Co
mp
lex
Gen
es
Str
ain
acs
Fb
chB
bch
Cb
chF
bch
Gb
chI
bch
Lb
chM
bch
Nb
chP
bch
Xb
chY
bch
Zcr
tAcr
tBcr
tCcr
tDcr
tEcr
tFcr
tIcr
tKp
ucC
pu
fAp
ufB
pu
fCp
ufL
pu
fMp
uh
A
Rh
izo
biu
msp
W
L3
Ro
seo
mo
nas
sp
WL4
5
Als
ob
act
er
sp
WL4
Gro
up
1M
eth
ylo
bact
eri
um
sp
WL9
Meth
ylo
bact
eri
um
sp
WL1
9
Meth
ylo
bact
eri
um
sp
WL6
9
Gro
up
2M
eth
ylo
bact
eri
um
sp
WL1
Meth
ylo
bact
eri
um
sp
WL2
Meth
ylo
bact
eri
um
sp
WL7
Meth
ylo
bact
eri
um
sp
WL1
8
Meth
ylo
bact
eri
um
sp
WL6
4
Gro
up
3M
eth
ylo
bact
eri
um
sp
WL6
Meth
ylo
bact
eri
um
sp
WL3
0
Meth
ylo
bact
eri
um
sp
WL9
3
Meth
ylo
bact
eri
um
sp
WL1
16
Meth
ylo
bact
eri
um
sp
WL1
19
Gro
up
4M
eth
ylo
bact
eri
um
sp
WL8
Meth
ylo
bact
eri
um
sp
WL1
2
Meth
ylo
bact
eri
um
sp
WL1
03
Meth
ylo
bact
eri
um
sp
WL1
20
Meth
ylo
bact
eri
um
sp
WL1
22
NO
TEmdash
Gra
yb
oxe
sin
dic
ate
pre
sen
ceo
fg
en
es
an
dw
hit
eb
oxe
sin
dic
ate
ab
sen
ceo
fg
en
es
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2903
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
of these two groups rely on the zeaxanthin pathway to pro-
duce carotenoid pigments and most likely nostoxanthin and
erythroxanthin which however do not participate in the
light-harvesting process but rather have a photoprotection
role (Noguchi et al 1992 Yurkov and Csotonyi 2009)
Phylogenetic Observations
In our study we sequenced and analyzed 21 aerobic anoxy-
genic photosynthetic bacteria most of which belonged to the
genus Methylobacterium the presence of which in the
environment has already been discussed previously
Methylobacterium spp are facultative methylotrophs and
can utilize a variety of C1 C2 C3 and C4 compounds
Such compounds are often found in the phyllosphere due
to plant metabolism (Vorholt 2012 Remus-Emsermann
et al 2014) Methylobacterium spp show a wide range in
chromosome size and number of plasmids they contain
(Marx et al 2012) The genus is represented well in
GenBank with 100 complete or draft genomes (last accessed
April 2019) However most of the sequences do not have
FIG 5mdashArchitecture of the different photosynthetic gene clusters Arrows indicate the orientation of the genes Sizes are not representatives of gene
length Only the genes shared by all PGCs were shown Genes that may be missing from some strains are denoted with parentheses on the figurersquos legend
Table 3
General Descriptive Statistics of the Four Complete AAP Bacterial Genomes
Taxon Genome Size (bp) GC Number of Genes Plasmids Plasmids Size (bp) GC
Rhizobium sp WL3 4568855 6160 4450 1 700631 586
mdash mdash mdash 2 85350 575
Roseomonas sp WL45 4533887 708 5042 1 262815 655
1011854 707 1138 2 240556 662
mdash mdash mdash 3 152922 658
mdash mdash mdash 4 85279 646
mdash mdash mdash 5 84774 661
mdash mdash mdash 6 83273 656
mdash mdash mdash 7 65860 648
Alsobacter sp WL4 5405668 679 5013 1 27178 620
mdash mdash mdash 2 9451 606
Methylobacterium sp WL1 6215463 691 6431 1 35731 638
Zervas et al GBE
2904 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
species affiliation (fig 4) Thus we were not able to assign any
of the 18 isolates to specific Methylobacterium species apart
perhaps from group 2 (WL1 WL2 WL7 WL18 WL64) which
is placed as sister to M brachiatum in the MASH-ANI analysis
Even in this case though M pseudosasicola BL36 is also
placed in the same clade as M brachiatum which raises the
question of proper nomenclature This is further evidenced in
the same tree (cyanide clade below) where seven species
share gt90 of their genome For the remaining isolates
WL19 is closely affiliated to other Methylobacterium strains
that have not been identified A similar picture is shown for
WL69 On the other hand WL19 appears to be a novel spe-
cies with unique genomic features and so do the strains from
groups 3 and 4 (red clade) Thus despite how ubiquitous
Methylobacterium is in nature it is clear that part of its diver-
gence is still missing Moreover a critical evaluation of the
current nomenclature of the genus would be beneficial based
on the whole genome sequences because closely related
genomes (MASH-ANI identitygt 97) have been given differ-
ent species affiliations
Apart from this genus we also identified a Rhizobium sp
a Roseomonas sp and a novel Alsobacter strain (WL4) that
belongs in the Methylocystaceae a family of methanotroph
bacteria This is the first time that AAP bacteria from this
family have been isolated and their complete genomes assem-
bled Its closely related Alsobacter spp have been shown to
exhibit high Thallium tolerance (Bao et al 2006) and it will be
interesting to identify heavy-metal resistance genes in non-
heavy-metal polluted phyllosphere samples It is worth men-
tioning that if WL4 contains heavy-metal resistance genes on
top of the PGC this will have been shown for the first time
and its potential to bioremediate heavy-metal contaminated
areas if applied as plant growth promotion and environmental
remediation agent is of high value and needs to further be
explored
Molecular Evolution of the PGCs
For all gene trees of the PGC we observed that Rhizobium sp
WL3 evolves faster than the other isolates followed by
Table 4
Substitutions Per Site in the Different Phylogenetic Analyses
Strain 16S acsF bchL bchY crtB pufL pufMsuper-matrix
CheckM
Roseomonas sp WL45 0096 0147 0100 0235 0250 0095 0132 0160 0184
Rhizobium sp WL3 0184 0622 0767 0690 1370 0343 0525 0580 0372
Alsobacter sp WL4 0151 0462 0233 0490 0870 0258 0334 0380 0403
Methylobacterium sp WL9 0226 0418 0372 0420 0670 0270 0319 0364 0477
Methylobacterium sp WL19 0221 0413 0300 0460 0600 0233 0346 0364 0468
Methylobacterium sp WL69 0231 0373 0333 0470 0620 0260 0358 0368 0468
Methylobacterium sp WL1 0233 0422 0344 0485 0480 0280 0385 0380 0459
Methylobacterium sp WL2 0233 0422 0344 0485 0480 0280 0385 0380 0459
Methylobacterium sp WL7 0238 0427 0344 0480 0480 0278 0385 0380 0459
Methylobacterium sp WL18 0238 0427 0344 0480 0480 0280 0385 0380 0459
Methylobacterium sp WL64 0233 0422 0344 0480 0485 0280 0385 0380 0459
Methylobacterium sp WL6 0199 0378 0256 0425 0680 0208 0311 0340 0455
Methylobacterium sp WL30 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL93 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL116 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL119 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL8 0188 0373 0258 0430 0680 0230 0358 0348 0459
Methylobacterium sp WL12 0188 0373 0256 0430 0680 0230 0358 0348 0459
Methylobacterium sp WL103 0188 0373 0256 0430 0850 0230 0358 0364 0459
Methylobacterium sp WL120 0188 0373 0258 0430 0680 0233 0358 0348 0459
Methylobacterium sp WL122 0188 0373 0256 0430 0680 0230 0358 0348 0459
Susbstitutions per site
Gro
up1
Gro
up 2
Gro
up 3
Gro
up 4
NOTEmdashAll values correspond to distances from the root where Roseomonas sp WL45 was chosen as outgroup for all the alignments
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2905
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Alsobacter sp WL4 Both these isolates also show much
higher substitution rates compared with their respective 16S
genes and core proteomes (checkM) which indicates that the
PGCs are under relaxed selection pressure In all instances the
phylogenetic trees constructed showed the same topology
and the same grouping of the Methylobacterium isolates
which suggests that the PGCs were incorporated in the
genomes of these strains relatively early in their evolution If
the PGC was free to move laterally then we would have ob-
served very different phylogenetic trees for its genes com-
pared with the 16S and core proteome analyses
Interestingly these observations do not reflect on the archi-
tecture of the PGC where Roseomonas sp WL45 and
Rhizobium sp WL3 contain the PGC in one continuous re-
gion whereas Alsobacter sp WL4 and Methylobacterium sp
WL1 have it in two pieces
For the 18 Methylobacterium isolates group 1 and 2
strains (WL9 WL19 WL69 and WL1 WL2 WL7 WL18
WL64) evolve consistently at different rates compared
with the groups 3 and 4 It appears that there is relatively
strong selection pressure on bchL pufL and pufM which
in all cases are more similar to each other and also show
slower molecular evolution rates compared with ascF
bchY and crtB which in all Methylobacterium isolates
evolve at least twice as fast compared with the 16S
gene (table 4) They also show fewer substitutions per
site compared with the core proteome analysis This result
partially contrasted with previously adopted methods of
using bchY as a marker gene for RC1 clusters (eg
Atamna-Ismaeel and Finkel 2012a) On the other hand
the use of pufM appears more sensible It would how-
ever be more suitable to investigate the possibility of us-
ing other genes for metagenome amplicon sequencing
approaches such as pufL or bchL which show a smaller
degree of divergence in different genera thus universal
primers might be more successful in detecting a wide va-
riety of AAP bacteria in environmental samples
Interestingly while Groups 1 and 2 isolates exhibit higher
substitution rates of the PGC genes compared with
Groups 3 and 4 this is not the case for crtB which evolves
a lot slower and also pufM which shows similar molec-
ular evolution in the genus Overall it seems that the
PGCs were incorporated in ancestral strains that later di-
verged to give rise to the different AAP taxa This is fur-
ther corroborated by the fact that almost all phylogenetic
trees (eg fig 5) are consistent and in agreement with
both the 16S phylogenetic tree (fig 1) as well as the phy-
logenomic tree (fig 2) If the genes of the PGCs had been
transferred to these AAP several times there would have
been significant differences both in the sequence level
and the phylogenetic analysis of the different genes
which would be expected to result in different topologies
and diverse distances from the root even for closelymdash
based on 16S analysismdashrelated strains
The Significance of AAP Bacteria in Wheat Phyllosphere
In this study we isolated for the first time a variety of AAP
bacteria that are present in the wheat phyllosphere adding to
the metagenomics knowledge provided earlier from five other
plants (Atamna-Ismaeel and Finkel 2012a 2012b) The phyllo-
sphere is a rather harsh environment with little water avail-
ability scarce resources very few suitable locations extreme
temperature difference between day and night and high
dosages of UV radiation (Vorholt 2012) AAP bacteria have
the ability to utilize light in photophosphorylation processes
through which they produce ATP and at the same time funnel
the few available nutrients in other metabolic pathways
(Yurkov and Hughes 2017) This has been shown to give
them an edge in vitro after exposure to light compared with
nonAAP bacteria (Ferrera et al 2011) Considering that phyl-
losphere bacteria protect the plant from pathogenic microbes
by secreting metabolites and by occupying the available
niches being able to persevere and outgrow other bacteria
gives a significant edge to AAP taxa and probably points
toward the suggestion of positive selection of such bacteria
in their phyllosphere by the plants themselves as they appear
to be absent from the soil (Atamna-Ismaeel and Finkel
2012b) At this point it is not clear exactly how AAP bacteria
benefit the plant especially since the phyllosphere is a com-
plex habitat with an ever-changing harsh environment In the
past years great efforts have been carried out on how to
improve crop production by studying the rhizosphere It has
also been suggested that the phyllosphere communities play a
significant role in plant health and growth which ultimately
leads to increased yields (Lindow and Brandl 2003) Given the
current projections of world population growth and the sup-
ply of food (UN 2017) it becomes evident that more in-depth
studies investigating the phyllosphere of plants and especially
crops and identifying the underlying drivers of the bacterial
communities and their interactions with their plant hosts is of
paramount importance Thus by isolating maintaining and
analyzing the genomes of phyllosphere isolates we may iden-
tify strains that may prove useful as plant growth promoting
agents in future field trials investigating the effect of phyllo-
sphere inoculation on plant growth health and yield
Supplementary Material
Supplementary data are available at Genome Biology and
Evolution online
Acknowledgments
The authors would like to thank Tue Sparholt Jorgensen for
collecting the wheat sample The authors also want to thank
Tina Thane (AU) Tanja Begovic (AU) and Margit Frederiksen
(NRCWE) for capable laboratory assistance We thank Michal
Koblızek for providing the initial model of the colony infrared
imaging system This work was funded by a Villum
Zervas et al GBE
2906 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Experiment grant (no 17601) and a Marie Skłodowska-Curie
AIAS-COFUND fellowship (EU-FP7 program Grant
Agreement No 609033) awarded to YZ
Author Contributions
AZ YZ LHH designed the study YZ isolated the strains
performed colony infrared imaging and extracted DNA for
Illumina sequencing YZ and AMM performed the MALDI-
TOF MS analysis AZ extracted DNA and performed ONT
Sequencing AZ carried out the bioinformatics and phyloge-
netic analyses AZ and YZ drafted the manuscript AZ
YZ AMM and LHH revised the manuscript for
intellectual content All authors read and approved the
manuscript
Literature Cited
Atamna-Ismaeel N Finkel O 2012a Bacterial anoxygenic photosynthesis
on plant leaf surfaces Environ Microbiol Rep 4(2)209ndash216
Atamna-Ismaeel N Finkel OM 2012b Microbial rhodopsins on leaf sur-
faces of terrestrial plants Environ Microbiol 14(1)140ndash146
Aziz RK et al 2008 The RAST Server rapid annotations using subsystems
technology BMC Genomics 9(1)75
Bankevich A et al 2012 SPAdes a new genome assembly algorithm and
its applications to single-cell sequencing J Comput Biol
19(5)455ndash477
Bao Z Sato Y Kubota M Ohta H 2006 Isolation and characterization of
thallium-tolerant bacteria from heavy metal-polluted river sediment
and non-polluted soils Microb Environ 21(4)251ndash260
Beatie GA 2002 Leaf surface waxes and the process of leaf colonization
by microorganisms In Lindow SE Hecht-Poinar EI Elliott VJ editors
Phyllosphere Microbiology St Paul (USA) APS Press pp 3ndash26
Beattie GA Lindow SE 1999 Bacterial colonization of leaves a spectrum
of strategies Phytopathology 89(5)353ndash359
Brettin T et al 2015 RASTtk a modular and extensible implementation of
the RAST algorithm for building custom annotation pipelines and an-
notating batches of genomes Sci Rep 58365
Carvalho SD Castillo JA 2018 Influence of light on plantndashphyllosphere
interaction Front Plant Sci 91482
Ferrera I Gasol JM Sebastian M Hojerova E Koblızek M 2011
Comparison of growth rates of aerobic anoxygenic phototrophic bac-
teria and other bacterioplankton groups in coastal Mediterranean wa-
ters Appl Environ Microbiol 77(21)7451ndash7458
Ferrera I Sanchez O Kolarova E Koblızek M Gasol JM 2017 Light
enhances the growth rates of natural populations of aerobic anoxy-
genic phototrophic bacteria ISME J 11(10)2391ndash2393
Gorlach K Shingaki R Morisaki H Hattori T 1994 Construction of eco-
collection of paddy field soil bacteria for population analysis J Gen
Appl Microbiol 40(6)509ndash517
Gourion B Rossignol M Vorholt JA Lindow SE 2006 A proteomic study
of Methylobacterium extorquens reveals a response regulator essential
for epiphytic growth wwwintegratedgenomicscom Accessed
November 26 2018
Green PN 2006 Methylobacterium In Dworkin M Falkow S Rosenberg
E Schleifer K Stackebrandt E editors The prokaryotes ndash a handbook
on the biology of bacteria New York Springer p 257ndash265
Hauruseu D Koblızek M 2012 Influence of light on carbon utilization in
aerobic anoxygenic phototrophs Appl Environ Microbiol
78(20)7414ndash7419
Katoh K Misawa K Kuma K Miyata T 2002 MAFFT a novel method for
rapid multiple sequence alignment based on fast Fourier transform
Nucleic Acids Res 30(14)3059ndash3066
Knief C et al 2012 Metaproteogenomic analysis of microbial communi-
ties in the phyllosphere and rhizosphere of rice ISME J
6(7)1378ndash1390
Knief C Frances L Vorholt JA 2010 Competitiveness of diverse methyl-
obacterium strains in the phyllosphere of arabidopsis thaliana and
identification of representative models including M extorquens
PA1 Microb Ecol 60(2)440ndash452
Kwak MJ et al 2014 Genome information of Methylobacterium oryzae a
plantndashprobiotic methylotroph in the phyllosphere PLoS One
9(9)e106704
Lindow SE Brandl MT 2003 Microbiology of the phyllosphere Appl
Environ Microbiol 69(4)1875ndash1883
Madsen AM Zervas A Tendal K Nielsen JL 2015 Microbial diversity in
bioaerosol samples causing ODTS compared to reference bioaerosol
samples as measured using Illumina sequencing and MALDI-TOF
Environ Res 140255ndash267
Martin M 2011 Cutadapt removes adapter sequences from high-
throughput sequencing reads EMBnet J 17(1)10ndash12
Marx CJ et al 2012 Complete genome sequences of six strains of the
genus Methylobacterium b J Bacteriol 194(17)4746
Noguchi T Hayashi H Shimada K Takaichi S Tasumi M 1992 In vivo
states and functions of carotenoids in an aerobic photosynthetic bac-
terium Erythrobacter longus Photosynth Res 31(1)21ndash30
Olm MR Brown CT Brooks B Banfield JF 2017 dRep a tool for fast and
accurate genomic comparisons that enables improved genome recov-
ery from metagenomes through de-replication ISME J
11(12)2864ndash2868
Overbeek R et al 2014 The SEED and the Rapid Annotation of microbial
genomes using Subsystems Technology (RAST) Nucl Acids Res
42(D1)D206ndashD214
Parks DH Imelfort M Skennerton CT Hugenholtz P Tyson GW 2015
CheckM assessing the quality of microbial genomes recovered from
isolates single cells and metagenomes Genome Res 25(7)1043ndash1055
Remus-Emsermann MNP et al 2014 Spatial distribution analyses of nat-
ural phyllosphere-colonizing bacteria on Arabidopsis thaliana revealed
by fluorescence in situ hybridization Environ Microbiol
16(7)2329ndash2340
Stamatakis A 2006 RAxML-VI-HPC maximum likelihood-based phyloge-
netic analyses with thousands of taxa and mixed models
Bioinformatics 22(21)2688ndash2690
UN 2017 World population prospects the 2017 revision Data Booklet
httpspopulationunorgwppPublicationsFilesWPP2019_DataBook-
letpdf
Vesth T Lagesen K Acar euroO Ussery D 2013 CMG-biotools a free work-
bench for basic comparative microbial genomics PLoS One
8(4)e60120
Vorholt JA 2012 Microbial life in the phyllosphere Nat Rev Microbiol
10(12)828ndash840
Wick RR Judd LM Gorrie CL Holt KE 2017 Unicycler resolving bacterial
genome assemblies from short and long sequencing reads PLoS
Comput Biol 13(6)e1005595
Wilson M Lindow SE 1994a Coexistence among epiphytic bacterial pop-
ulations mediated through nutritional resource partitioning Appl
Environ Microbiol 604468ndash77
Wilson M Lindow SE 1994b Ecological similarity and coexistence of epi-
phytic ice-nucleating (ice) pseudomonas syringae strains and a non-
ice-nucleating (ice) biological control agent Appl Environ Microbiol
60(9)3128ndash3137
Woodward FI Lomas MR 2004 Vegetation dynamics ndash simulating
responses to climatic change Biol Rev 79(3)643ndash670
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2907
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Xu J Gordon JI 2003 Honor thy symbionts Proc Natl Acad Sci USA
100(18)10452ndash10459
Yurkov V Csotonyi JT 2009 New light on aerobic anoxygenic photo-
trophs In Hunter NC Daldal F Thurnauer MC Beatty JT editors
The purple photosynthetic bacteria Dordrecht Springer thorn Business
Media BV p 31ndash55 doi101007978-1-4020-8815-5_3
Yurkov V Hughes E 2017 Aerobic Anoxygenic Phototrophs four
decades of mystery In Hallenbeck PC editor Modern topics in the
phototrophic prokaryotes environmental and applied aspects p
193ndash214
Yutin N et al 2009 BchY-based degenerate primers target all types of
anoxygenic photosynthetic bacteria in a single PCR Appl Environ
Microbiol 75(23)7556ndash7559
Zeng Y Feng F Medova H Dean J Koblizek M 2014 Functional
type 2 photosynthetic reaction centers found in the rare bacterial phy-
lum Gemmatimonadetes Proc Natl Acad Sci USA 111(21)7795ndash7800
Zheng Q et al 2011 Diverse arrangement of photosynthetic gene
clusters in aerobic anoxygenic phototrophic bacteria PLoS One
6(9)e25050
Associate editor Esperanza Martinez-Romero
Zervas et al GBE
2908 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
- evz204-TF1
- evz204-TF2
-
characters created with CheckM (Parks et al 2015) and its
lineage_wf algorithm with default settings Substitution rates
were calculated as distances from the root using the legend
provided by the phylogenetic trees
Results
Identification and Isolation of AAP Bacteria from thePhyllosphere
The 129 phyllosphere isolates that were tested positive for the
presence of bacteriochlorophyll-a were placed in 21 groups
according to their protein mass profiles based on MALDI-TOF
MS (supplementary fig MALDI-TOF Supplementary Material
online) A representative from each group was sequenced
(Illumina) and their genomes assembled (SPAdes) WL1 and
WL2 were from the same MALDI-TOF group and their
genomes are identical (ANI 100) confirming the validity
of MALDI-TOF MS for grouping similar isolates Information
about their phylogeny total genome size and genome as-
sembly statistics are presented in table 1
Analysis of the 16S genes showed that 17 isolates
belonged in Methylobacterium (separated in four distinct
groups) whereas the remaining three belonged to
Rhizobium Roseomonas and one novel species was affiliated
to the Methylocystaceae family (WL4) with most hits belong-
ing to uncultured environmental strains The only character-
ized hits of WL4 belonged to Methylosinus trichosporium
(Gorlach et al 1994) and Alsobacter metallidurans a recently
characterized novel genus novel species (Bao et al 2006)
However in both instances the similarity was low
(985) We included strains of both species and created
a 16S phylogenetic tree (fig 1) which placed Alsobacter sp
WL4 as sister to the Alsobacter-Methylosinus clade
The core proteome analysis (fig 2mdashCheckM) including M
trichosporium OB3b and Alsobacter sp SH9 established the
phylogeny of strain WL4 and it was highly supported as sister
to Alsobacter sp SH9 (BSfrac14 100) both of which were
distinct from M trichosporium OB3b (BSfrac14 100)
Methylobacterium strains were still placed into four distinct
groups (Group 1 WL9 WL19 WL69 Group 2 WL1 WL2
WL7 WL18 WL64 Group 3 WL6 WL30 WL93 WL116
WL119 Group 4 WL8 WL12 WL103 WL120) This was
further supported by the whole proteome comparison
(fig 3mdashBLAST matrix) For Groups 3 and 4 the average
core protein content is 4500 proteins whereas for Group
2 it is close to 4000 This may signify the presence of
either more or larger plasmids that characterize the strains
of this group For Group 1 this number falls even more as
in this case there clearly are three distinct species that make
up the group
We expanded the analysis to include 100 complete and
draft Methylobacterium spp genomes available in GenBank
(last assessed April 2019) and also other representatives from
the Methylocystaceae aiming at further establishing strain
WL4 in the genus Alsobacter and also providing greater res-
olution on the Methylobacterium isolates with a special focus
on the divergent group 1 Information about the included
genomes is given in supplementary table 1 Supplementary
Material online We performed whole genome comparisons
using ANI (fig 4mdashdRep)
The analysis showed that Group 1 isolates belonged in
different Methylobacterium species groups 3 and 4 were dif-
ferent strains of the same species (MASH ANI identity be-
tween groupsgt 90) whereas group 3 isolates are closely
related strains possibly belonging to Methylobacterium bra-
chiatum Strain WL4 is again placed close to the two identical
Alsobacter genomes (Alsobacter sp SH9 and A metallidurans
PVZS01) far away from the three remaining species from
within the Methylocystaceae (M trichosporium OB3b
Methylocystis rosea GW6 Methylocystis bryophila S285)
However the MASH-ANI identity between WL4 and
Alsobacter is below the 80 threshold of genus assignment
Therefore we took the top 200 megablast hits for WL4rsquos 16S
region (information provided in supplementary table 2
Supplementary Material online) aligned them including strain
WL4 as mentioned previously and created a phylogenetic tree
of the 201 sequences (supplementary fig 1 Supplementary
Material online) The results clearly show that WL4 belongs in
Alsobacter whereas Methylosinus spp form their own dis-
tinct cluster
Table 1
Genome Statistics and Phylogenetic Information of the 21 Isolates that
were Illumina Sequenced in This Study
Strain Taxonomy Contigs Genome Size N50 GC
WL1 Methylobacterium 654 6258106 19384 6892
WL2 Methylobacterium 298 6292310 52057 6899
WL3 Rhizobium 55 5333961 210875 6114
WL4 Alsobacter 139 5412272 86934 6792
WL6 Methylobacterium 708 5549874 15034 6972
WL7 Methylobacterium 431 6059272 36478 6904
WL8 Methylobacterium 225 5322599 56100 6969
WL9 Methylobacterium 221 4670253 42885 6741
WL12 Methylobacterium 271 5545758 53038 694
WL18 Methylobacterium 1373 5796709 7880 6864
WL19 Methylobacterium 330 5177712 31633 6698
WL30 Methylobacterium 377 5636306 38077 6978
WL45 Roseomonas 2056 5961800 5511 6954
WL64 Methylobacterium 365 6908402 48840 6834
WL69 Methylobacterium 225 4693485 43639 6951
WL93 Methylobacterium 719 5642324 16902 6945
WL103 Methylobacterium 1327 5167285 7154 6897
WL116 Methylobacterium 1173 5445855 8413 6943
WL119 Methylobacterium 263 5709790 45485 6958
WL120 Methylobacterium 395 5266148 26698 6954
WL122 Methylobacterium 2024 4508072 3666 6877
Zervas et al GBE
2898 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Molecular Evolution of the PGC
The PGCs identified in the sequenced isolates all belong to
RC-II type meaning they contain bacteriochlorophyll-a and
possess pheophytinmdashquinone type reaction centers All
strains possess a full bacteriochlorophyll synthase gene set
except for Alsobacter sp WL4 that is lacking the bchM
gene (table 2)
For the carotenoid biosynthesis genes all strains possess
crtB whereas crtA is only present in Alsobacter sp WL4 and
Rhizobium sp strain WL3 For the crtC and crtK genes an
interesting pattern is observed Methylobacterium group 1
and group 2 strains contain crtC (with three exceptions dis-
cussed below) whereas group 3 and 4 strains completely lack
the gene On the other hand crtK is only present in
FIG 1mdashPhylogenetic tree of the 16S genes of the sequenced isolates Methylosinus trichosporium and Alsobacter metallidurans were included to better
pinpoint the phylogenetic placement of strain WL4 The legend shows substitution rates from the root of the tree and the node labels indicate bootstrap
support values ()
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2899
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Methylobacterium group 3 strains and missing from groups 1
2 and 4 Similar patterns are evidenced in pufB which is ab-
sent from Methylobacterium groups 3 and 4 Rhizobium sp
WL3 and Roseomonas sp WL45 and pufC missing from
Methylobacterium group 4 as well as the novel strain WL4
The remaining genes were found ubiquitously in all strains of
the analysis with only a few exceptions (table 2)
The PGCs were consistently found on the longer contigs of
the initial assemblies of Illumina data produced by SPAdes
(for the strains with fewer than 500 contigs) (table 1) In
Rhizobium sp WL3 and Roseomonas sp WL45 the photosyn-
thetic genes form one continuous cluster with different
however architectures In Alsobacter sp WL4 the genes are
organized in two clusters similar to the Methylobacterium
spp isolates In these cases the genes were found in the
middle of the contigs ruling out the possibility of PGC being
organized into two clusters as an artifact of the assembly
Unique to the architecture present in WL4 is the presence
FIG 2mdashPhylogenomic tree based on the core proteome (CheckM) of the 21 pure bacterial isolates of the analysis including Methylosinus trichosporium
OB3B and Alsobacter sp SH9 The legend shows substitution rates from the root of the tree and the node labels indicated bootstrap support values ()
Zervas et al GBE
2900 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
of bchI before the crtB crtI genes in the first cluster For
Methylobacterium spp groups 2 and 3 show similar gene
syntenymdashthough there are a few differences in gene content
(table 2)mdashwhereas groups 1 and 4 appear to also share the
overall architecture of the PGCs For isolates WL9 and WL19
(both in group 1) the exact location of the crtB and crtI genes
is not known as they appear on a short contig by themselves
Annotating the assemblies on RAST revealed several house-
keeping genes present in the vicinity of these clusters which
further supported the notion of their chromosomal place-
ment The completed genomes resulting from the hybrid as-
semblies (strains WL1 WL3 WL4 and WL45) provided the
final evidence of the exact placement of the PGCs on
the chromosomes of all our isolates The different
architectures of the isolated PGCs are shown in figure 5
General descriptive statistics of the assembled complete
genomes are given in table 3 The hybrid assemblies resulted
in both complete genomes and plasmids Thus Rhizobium sp
WL3 contains two plasmids Roseomonas sp WL45 contains
two chromosomes and seven plasmids Alsobacter sp WL4
with two plasmids Methylobacterium sp WL1 with one
plasmid
Using Roseomonas sp WL45 as outgroup we generated
phylogenetic trees for the 16S acsF bchL bchY crtB pufL
and pufM genes as described previously (data not shown)
We calculated substitution rates from the root of the different
trees including the core proteome (CheckM) tree and com-
pared them to each other (table 4)
Discussion
A New Strategy to Generate Complete Genomes ofUnique Aerobic Anoxygenic Phototrophic Bacteria FromEnvironmental Isolates
Aiming to enhance our knowledge of the prevalence of aer-
obic anoxygenic phototrophic bacteria in the phyllosphere
we designed this study using wheat as the plant of choice
and employed a high-throughput multi-disciplinary approach
to identify isolate and analyze AAP strains From the initial
137 plates we ended up with 4480 bacterial colonies
gt500 of which were probable AAP strains After macroscopic
investigation of size shape color and texture of the colonies
we chose 129 strains for which we ran proteome profiling
using MALDI-TOF MS This approach has been shown to be
FIG 3mdashPairwise proteome comparisons (BLAST-matrix) of the 21 bacterial isolates of the analysis The gray-to-green color gradient indicates low to
high percentages of identity The bottom boxes show the presence of homologs within the same proteome (gray-to-red color gradient shows low to high
presence of homologs in the specific proteome)
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2901
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
able to reproducibly distinguish different taxa and in cases
where reference MALDI-TOF spectra are available in
the database be able to assign Genus andor Species names
(Madsen et al 2015) We then proceeded with whole ge-
nome sequencing on the Illumina Nextseq platform and for
the unique genomes we also proceeded with Oxford
Nanopore Technologies sequencing to generate complete
genomes and plasmids
Following this multi-disciplinary approach that relies on
classical microbiological techniques (plating morphology
etc) biochemical analyses (MALDI-TOF MS) and two differ-
ent types of high-throughput whole genome sequencing we
were able to generate draft and complete genomes of novel
AAP bacteria from the wheat phyllosphere in a relatively quick
(weeks) and efficient manner without losing novel biological
information in the process of reducing redundancy
Bacteriochlorophyll-Containing AAP Bacteria in the WheatPhyllosphere
By investigating the presence of bacteriochlorophylls in the
wheat phyllosphere we were able to identify 21 unique strains
harboring them the majority of which belonged to the genus
Methylobacterium Members of this genus are ubiquitous in
nature and have been detected in both soil and plant surfaces
(Green 2006) Targeting the bchY (chlorophyllide reductase
subunit Y) and pufM (reaction centre subunit M) genes in
phyllosphere samples of different plants it was shown that
the Methylobacteriaceae (alpha-Proteobacteria) comprised
28 of the total pufM sequences from the amplicon sequenc-
ing analysis (Atamna-Ismaeel and Finkel 2012a) Our results
are thus in accordance to these previous studies As
Methylobacteriaceae are indigenous inhabitants of the phyllo-
sphere and have consistently been shown to harbor PGCs it
may be that these photosystems are essential to the
Methylobacteria that inhabit this niche Interestingly in their
study Atamna-Ismaeel et al had negative results from their
bchY amplicon sequencing approach suggesting the absence
of bacteria that contain RC1 (bacteriochlorophyll-a containing
reaction centers) type photosystems Our results show that all
isolates contain the complete cassette of chlorophyllide syn-
thase subunit encoding genes (bchB bchC bchF bchG bchI
bchL bchM bchN bchP bchX bchY bchZ) as well as both the
pufL and pufM genes (table 2) Testing the same primers (Yutin
et al 2009) in silico against our bchY sequences we had pos-
itive results for all of them (data not shown)
Patterns of loss of the carotenoid genes have been ob-
served in other alpha and gamma proteobacteria like in
Rhodobacterales and Sphingomonadales (Zheng et al
2011) The absence of crtA results in the inability to perform
the final step in the spheroidene pathway where hydroxy-
spheroidene is converted to hydroxyspheroidenone
Moreover the absence of crtC in Methylobacterium spp of
groups 3 and 4 completely affects both the spheroidene and
the spirilloxanthin pathways as its product is essential for the
first step of both metabolic pathways starting from neuro-
sporene and lycopene respectively This means that strains
FIG 4mdashWhole genome average nucleotide identity (ANI) cladogram
showing percentages of k-mer identities between the isolates Taxa with
the same color and above 90 identity threshold are considered to belong
in the same species Beijerinckia indica ATCC-9039 was used as outgroup
of the analyses (not defined a priori)
Zervas et al GBE
2902 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Tab
le2
Gen
eC
onte
nt
of
the
Photo
synth
esis
Gen
eC
lust
ers
Iden
tified
inth
e21
Bac
terial
Isola
tes
Bact
eri
och
loro
ph
yll
Syn
thase
Gen
es
Caro
ten
oid
Bio
syn
thesi
sG
en
es
Lig
ht-
Harv
est
ing
Co
mp
lex
Gen
es
Str
ain
acs
Fb
chB
bch
Cb
chF
bch
Gb
chI
bch
Lb
chM
bch
Nb
chP
bch
Xb
chY
bch
Zcr
tAcr
tBcr
tCcr
tDcr
tEcr
tFcr
tIcr
tKp
ucC
pu
fAp
ufB
pu
fCp
ufL
pu
fMp
uh
A
Rh
izo
biu
msp
W
L3
Ro
seo
mo
nas
sp
WL4
5
Als
ob
act
er
sp
WL4
Gro
up
1M
eth
ylo
bact
eri
um
sp
WL9
Meth
ylo
bact
eri
um
sp
WL1
9
Meth
ylo
bact
eri
um
sp
WL6
9
Gro
up
2M
eth
ylo
bact
eri
um
sp
WL1
Meth
ylo
bact
eri
um
sp
WL2
Meth
ylo
bact
eri
um
sp
WL7
Meth
ylo
bact
eri
um
sp
WL1
8
Meth
ylo
bact
eri
um
sp
WL6
4
Gro
up
3M
eth
ylo
bact
eri
um
sp
WL6
Meth
ylo
bact
eri
um
sp
WL3
0
Meth
ylo
bact
eri
um
sp
WL9
3
Meth
ylo
bact
eri
um
sp
WL1
16
Meth
ylo
bact
eri
um
sp
WL1
19
Gro
up
4M
eth
ylo
bact
eri
um
sp
WL8
Meth
ylo
bact
eri
um
sp
WL1
2
Meth
ylo
bact
eri
um
sp
WL1
03
Meth
ylo
bact
eri
um
sp
WL1
20
Meth
ylo
bact
eri
um
sp
WL1
22
NO
TEmdash
Gra
yb
oxe
sin
dic
ate
pre
sen
ceo
fg
en
es
an
dw
hit
eb
oxe
sin
dic
ate
ab
sen
ceo
fg
en
es
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2903
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
of these two groups rely on the zeaxanthin pathway to pro-
duce carotenoid pigments and most likely nostoxanthin and
erythroxanthin which however do not participate in the
light-harvesting process but rather have a photoprotection
role (Noguchi et al 1992 Yurkov and Csotonyi 2009)
Phylogenetic Observations
In our study we sequenced and analyzed 21 aerobic anoxy-
genic photosynthetic bacteria most of which belonged to the
genus Methylobacterium the presence of which in the
environment has already been discussed previously
Methylobacterium spp are facultative methylotrophs and
can utilize a variety of C1 C2 C3 and C4 compounds
Such compounds are often found in the phyllosphere due
to plant metabolism (Vorholt 2012 Remus-Emsermann
et al 2014) Methylobacterium spp show a wide range in
chromosome size and number of plasmids they contain
(Marx et al 2012) The genus is represented well in
GenBank with 100 complete or draft genomes (last accessed
April 2019) However most of the sequences do not have
FIG 5mdashArchitecture of the different photosynthetic gene clusters Arrows indicate the orientation of the genes Sizes are not representatives of gene
length Only the genes shared by all PGCs were shown Genes that may be missing from some strains are denoted with parentheses on the figurersquos legend
Table 3
General Descriptive Statistics of the Four Complete AAP Bacterial Genomes
Taxon Genome Size (bp) GC Number of Genes Plasmids Plasmids Size (bp) GC
Rhizobium sp WL3 4568855 6160 4450 1 700631 586
mdash mdash mdash 2 85350 575
Roseomonas sp WL45 4533887 708 5042 1 262815 655
1011854 707 1138 2 240556 662
mdash mdash mdash 3 152922 658
mdash mdash mdash 4 85279 646
mdash mdash mdash 5 84774 661
mdash mdash mdash 6 83273 656
mdash mdash mdash 7 65860 648
Alsobacter sp WL4 5405668 679 5013 1 27178 620
mdash mdash mdash 2 9451 606
Methylobacterium sp WL1 6215463 691 6431 1 35731 638
Zervas et al GBE
2904 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
species affiliation (fig 4) Thus we were not able to assign any
of the 18 isolates to specific Methylobacterium species apart
perhaps from group 2 (WL1 WL2 WL7 WL18 WL64) which
is placed as sister to M brachiatum in the MASH-ANI analysis
Even in this case though M pseudosasicola BL36 is also
placed in the same clade as M brachiatum which raises the
question of proper nomenclature This is further evidenced in
the same tree (cyanide clade below) where seven species
share gt90 of their genome For the remaining isolates
WL19 is closely affiliated to other Methylobacterium strains
that have not been identified A similar picture is shown for
WL69 On the other hand WL19 appears to be a novel spe-
cies with unique genomic features and so do the strains from
groups 3 and 4 (red clade) Thus despite how ubiquitous
Methylobacterium is in nature it is clear that part of its diver-
gence is still missing Moreover a critical evaluation of the
current nomenclature of the genus would be beneficial based
on the whole genome sequences because closely related
genomes (MASH-ANI identitygt 97) have been given differ-
ent species affiliations
Apart from this genus we also identified a Rhizobium sp
a Roseomonas sp and a novel Alsobacter strain (WL4) that
belongs in the Methylocystaceae a family of methanotroph
bacteria This is the first time that AAP bacteria from this
family have been isolated and their complete genomes assem-
bled Its closely related Alsobacter spp have been shown to
exhibit high Thallium tolerance (Bao et al 2006) and it will be
interesting to identify heavy-metal resistance genes in non-
heavy-metal polluted phyllosphere samples It is worth men-
tioning that if WL4 contains heavy-metal resistance genes on
top of the PGC this will have been shown for the first time
and its potential to bioremediate heavy-metal contaminated
areas if applied as plant growth promotion and environmental
remediation agent is of high value and needs to further be
explored
Molecular Evolution of the PGCs
For all gene trees of the PGC we observed that Rhizobium sp
WL3 evolves faster than the other isolates followed by
Table 4
Substitutions Per Site in the Different Phylogenetic Analyses
Strain 16S acsF bchL bchY crtB pufL pufMsuper-matrix
CheckM
Roseomonas sp WL45 0096 0147 0100 0235 0250 0095 0132 0160 0184
Rhizobium sp WL3 0184 0622 0767 0690 1370 0343 0525 0580 0372
Alsobacter sp WL4 0151 0462 0233 0490 0870 0258 0334 0380 0403
Methylobacterium sp WL9 0226 0418 0372 0420 0670 0270 0319 0364 0477
Methylobacterium sp WL19 0221 0413 0300 0460 0600 0233 0346 0364 0468
Methylobacterium sp WL69 0231 0373 0333 0470 0620 0260 0358 0368 0468
Methylobacterium sp WL1 0233 0422 0344 0485 0480 0280 0385 0380 0459
Methylobacterium sp WL2 0233 0422 0344 0485 0480 0280 0385 0380 0459
Methylobacterium sp WL7 0238 0427 0344 0480 0480 0278 0385 0380 0459
Methylobacterium sp WL18 0238 0427 0344 0480 0480 0280 0385 0380 0459
Methylobacterium sp WL64 0233 0422 0344 0480 0485 0280 0385 0380 0459
Methylobacterium sp WL6 0199 0378 0256 0425 0680 0208 0311 0340 0455
Methylobacterium sp WL30 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL93 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL116 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL119 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL8 0188 0373 0258 0430 0680 0230 0358 0348 0459
Methylobacterium sp WL12 0188 0373 0256 0430 0680 0230 0358 0348 0459
Methylobacterium sp WL103 0188 0373 0256 0430 0850 0230 0358 0364 0459
Methylobacterium sp WL120 0188 0373 0258 0430 0680 0233 0358 0348 0459
Methylobacterium sp WL122 0188 0373 0256 0430 0680 0230 0358 0348 0459
Susbstitutions per site
Gro
up1
Gro
up 2
Gro
up 3
Gro
up 4
NOTEmdashAll values correspond to distances from the root where Roseomonas sp WL45 was chosen as outgroup for all the alignments
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2905
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Alsobacter sp WL4 Both these isolates also show much
higher substitution rates compared with their respective 16S
genes and core proteomes (checkM) which indicates that the
PGCs are under relaxed selection pressure In all instances the
phylogenetic trees constructed showed the same topology
and the same grouping of the Methylobacterium isolates
which suggests that the PGCs were incorporated in the
genomes of these strains relatively early in their evolution If
the PGC was free to move laterally then we would have ob-
served very different phylogenetic trees for its genes com-
pared with the 16S and core proteome analyses
Interestingly these observations do not reflect on the archi-
tecture of the PGC where Roseomonas sp WL45 and
Rhizobium sp WL3 contain the PGC in one continuous re-
gion whereas Alsobacter sp WL4 and Methylobacterium sp
WL1 have it in two pieces
For the 18 Methylobacterium isolates group 1 and 2
strains (WL9 WL19 WL69 and WL1 WL2 WL7 WL18
WL64) evolve consistently at different rates compared
with the groups 3 and 4 It appears that there is relatively
strong selection pressure on bchL pufL and pufM which
in all cases are more similar to each other and also show
slower molecular evolution rates compared with ascF
bchY and crtB which in all Methylobacterium isolates
evolve at least twice as fast compared with the 16S
gene (table 4) They also show fewer substitutions per
site compared with the core proteome analysis This result
partially contrasted with previously adopted methods of
using bchY as a marker gene for RC1 clusters (eg
Atamna-Ismaeel and Finkel 2012a) On the other hand
the use of pufM appears more sensible It would how-
ever be more suitable to investigate the possibility of us-
ing other genes for metagenome amplicon sequencing
approaches such as pufL or bchL which show a smaller
degree of divergence in different genera thus universal
primers might be more successful in detecting a wide va-
riety of AAP bacteria in environmental samples
Interestingly while Groups 1 and 2 isolates exhibit higher
substitution rates of the PGC genes compared with
Groups 3 and 4 this is not the case for crtB which evolves
a lot slower and also pufM which shows similar molec-
ular evolution in the genus Overall it seems that the
PGCs were incorporated in ancestral strains that later di-
verged to give rise to the different AAP taxa This is fur-
ther corroborated by the fact that almost all phylogenetic
trees (eg fig 5) are consistent and in agreement with
both the 16S phylogenetic tree (fig 1) as well as the phy-
logenomic tree (fig 2) If the genes of the PGCs had been
transferred to these AAP several times there would have
been significant differences both in the sequence level
and the phylogenetic analysis of the different genes
which would be expected to result in different topologies
and diverse distances from the root even for closelymdash
based on 16S analysismdashrelated strains
The Significance of AAP Bacteria in Wheat Phyllosphere
In this study we isolated for the first time a variety of AAP
bacteria that are present in the wheat phyllosphere adding to
the metagenomics knowledge provided earlier from five other
plants (Atamna-Ismaeel and Finkel 2012a 2012b) The phyllo-
sphere is a rather harsh environment with little water avail-
ability scarce resources very few suitable locations extreme
temperature difference between day and night and high
dosages of UV radiation (Vorholt 2012) AAP bacteria have
the ability to utilize light in photophosphorylation processes
through which they produce ATP and at the same time funnel
the few available nutrients in other metabolic pathways
(Yurkov and Hughes 2017) This has been shown to give
them an edge in vitro after exposure to light compared with
nonAAP bacteria (Ferrera et al 2011) Considering that phyl-
losphere bacteria protect the plant from pathogenic microbes
by secreting metabolites and by occupying the available
niches being able to persevere and outgrow other bacteria
gives a significant edge to AAP taxa and probably points
toward the suggestion of positive selection of such bacteria
in their phyllosphere by the plants themselves as they appear
to be absent from the soil (Atamna-Ismaeel and Finkel
2012b) At this point it is not clear exactly how AAP bacteria
benefit the plant especially since the phyllosphere is a com-
plex habitat with an ever-changing harsh environment In the
past years great efforts have been carried out on how to
improve crop production by studying the rhizosphere It has
also been suggested that the phyllosphere communities play a
significant role in plant health and growth which ultimately
leads to increased yields (Lindow and Brandl 2003) Given the
current projections of world population growth and the sup-
ply of food (UN 2017) it becomes evident that more in-depth
studies investigating the phyllosphere of plants and especially
crops and identifying the underlying drivers of the bacterial
communities and their interactions with their plant hosts is of
paramount importance Thus by isolating maintaining and
analyzing the genomes of phyllosphere isolates we may iden-
tify strains that may prove useful as plant growth promoting
agents in future field trials investigating the effect of phyllo-
sphere inoculation on plant growth health and yield
Supplementary Material
Supplementary data are available at Genome Biology and
Evolution online
Acknowledgments
The authors would like to thank Tue Sparholt Jorgensen for
collecting the wheat sample The authors also want to thank
Tina Thane (AU) Tanja Begovic (AU) and Margit Frederiksen
(NRCWE) for capable laboratory assistance We thank Michal
Koblızek for providing the initial model of the colony infrared
imaging system This work was funded by a Villum
Zervas et al GBE
2906 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Experiment grant (no 17601) and a Marie Skłodowska-Curie
AIAS-COFUND fellowship (EU-FP7 program Grant
Agreement No 609033) awarded to YZ
Author Contributions
AZ YZ LHH designed the study YZ isolated the strains
performed colony infrared imaging and extracted DNA for
Illumina sequencing YZ and AMM performed the MALDI-
TOF MS analysis AZ extracted DNA and performed ONT
Sequencing AZ carried out the bioinformatics and phyloge-
netic analyses AZ and YZ drafted the manuscript AZ
YZ AMM and LHH revised the manuscript for
intellectual content All authors read and approved the
manuscript
Literature Cited
Atamna-Ismaeel N Finkel O 2012a Bacterial anoxygenic photosynthesis
on plant leaf surfaces Environ Microbiol Rep 4(2)209ndash216
Atamna-Ismaeel N Finkel OM 2012b Microbial rhodopsins on leaf sur-
faces of terrestrial plants Environ Microbiol 14(1)140ndash146
Aziz RK et al 2008 The RAST Server rapid annotations using subsystems
technology BMC Genomics 9(1)75
Bankevich A et al 2012 SPAdes a new genome assembly algorithm and
its applications to single-cell sequencing J Comput Biol
19(5)455ndash477
Bao Z Sato Y Kubota M Ohta H 2006 Isolation and characterization of
thallium-tolerant bacteria from heavy metal-polluted river sediment
and non-polluted soils Microb Environ 21(4)251ndash260
Beatie GA 2002 Leaf surface waxes and the process of leaf colonization
by microorganisms In Lindow SE Hecht-Poinar EI Elliott VJ editors
Phyllosphere Microbiology St Paul (USA) APS Press pp 3ndash26
Beattie GA Lindow SE 1999 Bacterial colonization of leaves a spectrum
of strategies Phytopathology 89(5)353ndash359
Brettin T et al 2015 RASTtk a modular and extensible implementation of
the RAST algorithm for building custom annotation pipelines and an-
notating batches of genomes Sci Rep 58365
Carvalho SD Castillo JA 2018 Influence of light on plantndashphyllosphere
interaction Front Plant Sci 91482
Ferrera I Gasol JM Sebastian M Hojerova E Koblızek M 2011
Comparison of growth rates of aerobic anoxygenic phototrophic bac-
teria and other bacterioplankton groups in coastal Mediterranean wa-
ters Appl Environ Microbiol 77(21)7451ndash7458
Ferrera I Sanchez O Kolarova E Koblızek M Gasol JM 2017 Light
enhances the growth rates of natural populations of aerobic anoxy-
genic phototrophic bacteria ISME J 11(10)2391ndash2393
Gorlach K Shingaki R Morisaki H Hattori T 1994 Construction of eco-
collection of paddy field soil bacteria for population analysis J Gen
Appl Microbiol 40(6)509ndash517
Gourion B Rossignol M Vorholt JA Lindow SE 2006 A proteomic study
of Methylobacterium extorquens reveals a response regulator essential
for epiphytic growth wwwintegratedgenomicscom Accessed
November 26 2018
Green PN 2006 Methylobacterium In Dworkin M Falkow S Rosenberg
E Schleifer K Stackebrandt E editors The prokaryotes ndash a handbook
on the biology of bacteria New York Springer p 257ndash265
Hauruseu D Koblızek M 2012 Influence of light on carbon utilization in
aerobic anoxygenic phototrophs Appl Environ Microbiol
78(20)7414ndash7419
Katoh K Misawa K Kuma K Miyata T 2002 MAFFT a novel method for
rapid multiple sequence alignment based on fast Fourier transform
Nucleic Acids Res 30(14)3059ndash3066
Knief C et al 2012 Metaproteogenomic analysis of microbial communi-
ties in the phyllosphere and rhizosphere of rice ISME J
6(7)1378ndash1390
Knief C Frances L Vorholt JA 2010 Competitiveness of diverse methyl-
obacterium strains in the phyllosphere of arabidopsis thaliana and
identification of representative models including M extorquens
PA1 Microb Ecol 60(2)440ndash452
Kwak MJ et al 2014 Genome information of Methylobacterium oryzae a
plantndashprobiotic methylotroph in the phyllosphere PLoS One
9(9)e106704
Lindow SE Brandl MT 2003 Microbiology of the phyllosphere Appl
Environ Microbiol 69(4)1875ndash1883
Madsen AM Zervas A Tendal K Nielsen JL 2015 Microbial diversity in
bioaerosol samples causing ODTS compared to reference bioaerosol
samples as measured using Illumina sequencing and MALDI-TOF
Environ Res 140255ndash267
Martin M 2011 Cutadapt removes adapter sequences from high-
throughput sequencing reads EMBnet J 17(1)10ndash12
Marx CJ et al 2012 Complete genome sequences of six strains of the
genus Methylobacterium b J Bacteriol 194(17)4746
Noguchi T Hayashi H Shimada K Takaichi S Tasumi M 1992 In vivo
states and functions of carotenoids in an aerobic photosynthetic bac-
terium Erythrobacter longus Photosynth Res 31(1)21ndash30
Olm MR Brown CT Brooks B Banfield JF 2017 dRep a tool for fast and
accurate genomic comparisons that enables improved genome recov-
ery from metagenomes through de-replication ISME J
11(12)2864ndash2868
Overbeek R et al 2014 The SEED and the Rapid Annotation of microbial
genomes using Subsystems Technology (RAST) Nucl Acids Res
42(D1)D206ndashD214
Parks DH Imelfort M Skennerton CT Hugenholtz P Tyson GW 2015
CheckM assessing the quality of microbial genomes recovered from
isolates single cells and metagenomes Genome Res 25(7)1043ndash1055
Remus-Emsermann MNP et al 2014 Spatial distribution analyses of nat-
ural phyllosphere-colonizing bacteria on Arabidopsis thaliana revealed
by fluorescence in situ hybridization Environ Microbiol
16(7)2329ndash2340
Stamatakis A 2006 RAxML-VI-HPC maximum likelihood-based phyloge-
netic analyses with thousands of taxa and mixed models
Bioinformatics 22(21)2688ndash2690
UN 2017 World population prospects the 2017 revision Data Booklet
httpspopulationunorgwppPublicationsFilesWPP2019_DataBook-
letpdf
Vesth T Lagesen K Acar euroO Ussery D 2013 CMG-biotools a free work-
bench for basic comparative microbial genomics PLoS One
8(4)e60120
Vorholt JA 2012 Microbial life in the phyllosphere Nat Rev Microbiol
10(12)828ndash840
Wick RR Judd LM Gorrie CL Holt KE 2017 Unicycler resolving bacterial
genome assemblies from short and long sequencing reads PLoS
Comput Biol 13(6)e1005595
Wilson M Lindow SE 1994a Coexistence among epiphytic bacterial pop-
ulations mediated through nutritional resource partitioning Appl
Environ Microbiol 604468ndash77
Wilson M Lindow SE 1994b Ecological similarity and coexistence of epi-
phytic ice-nucleating (ice) pseudomonas syringae strains and a non-
ice-nucleating (ice) biological control agent Appl Environ Microbiol
60(9)3128ndash3137
Woodward FI Lomas MR 2004 Vegetation dynamics ndash simulating
responses to climatic change Biol Rev 79(3)643ndash670
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2907
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Xu J Gordon JI 2003 Honor thy symbionts Proc Natl Acad Sci USA
100(18)10452ndash10459
Yurkov V Csotonyi JT 2009 New light on aerobic anoxygenic photo-
trophs In Hunter NC Daldal F Thurnauer MC Beatty JT editors
The purple photosynthetic bacteria Dordrecht Springer thorn Business
Media BV p 31ndash55 doi101007978-1-4020-8815-5_3
Yurkov V Hughes E 2017 Aerobic Anoxygenic Phototrophs four
decades of mystery In Hallenbeck PC editor Modern topics in the
phototrophic prokaryotes environmental and applied aspects p
193ndash214
Yutin N et al 2009 BchY-based degenerate primers target all types of
anoxygenic photosynthetic bacteria in a single PCR Appl Environ
Microbiol 75(23)7556ndash7559
Zeng Y Feng F Medova H Dean J Koblizek M 2014 Functional
type 2 photosynthetic reaction centers found in the rare bacterial phy-
lum Gemmatimonadetes Proc Natl Acad Sci USA 111(21)7795ndash7800
Zheng Q et al 2011 Diverse arrangement of photosynthetic gene
clusters in aerobic anoxygenic phototrophic bacteria PLoS One
6(9)e25050
Associate editor Esperanza Martinez-Romero
Zervas et al GBE
2908 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
- evz204-TF1
- evz204-TF2
-
Molecular Evolution of the PGC
The PGCs identified in the sequenced isolates all belong to
RC-II type meaning they contain bacteriochlorophyll-a and
possess pheophytinmdashquinone type reaction centers All
strains possess a full bacteriochlorophyll synthase gene set
except for Alsobacter sp WL4 that is lacking the bchM
gene (table 2)
For the carotenoid biosynthesis genes all strains possess
crtB whereas crtA is only present in Alsobacter sp WL4 and
Rhizobium sp strain WL3 For the crtC and crtK genes an
interesting pattern is observed Methylobacterium group 1
and group 2 strains contain crtC (with three exceptions dis-
cussed below) whereas group 3 and 4 strains completely lack
the gene On the other hand crtK is only present in
FIG 1mdashPhylogenetic tree of the 16S genes of the sequenced isolates Methylosinus trichosporium and Alsobacter metallidurans were included to better
pinpoint the phylogenetic placement of strain WL4 The legend shows substitution rates from the root of the tree and the node labels indicate bootstrap
support values ()
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2899
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Methylobacterium group 3 strains and missing from groups 1
2 and 4 Similar patterns are evidenced in pufB which is ab-
sent from Methylobacterium groups 3 and 4 Rhizobium sp
WL3 and Roseomonas sp WL45 and pufC missing from
Methylobacterium group 4 as well as the novel strain WL4
The remaining genes were found ubiquitously in all strains of
the analysis with only a few exceptions (table 2)
The PGCs were consistently found on the longer contigs of
the initial assemblies of Illumina data produced by SPAdes
(for the strains with fewer than 500 contigs) (table 1) In
Rhizobium sp WL3 and Roseomonas sp WL45 the photosyn-
thetic genes form one continuous cluster with different
however architectures In Alsobacter sp WL4 the genes are
organized in two clusters similar to the Methylobacterium
spp isolates In these cases the genes were found in the
middle of the contigs ruling out the possibility of PGC being
organized into two clusters as an artifact of the assembly
Unique to the architecture present in WL4 is the presence
FIG 2mdashPhylogenomic tree based on the core proteome (CheckM) of the 21 pure bacterial isolates of the analysis including Methylosinus trichosporium
OB3B and Alsobacter sp SH9 The legend shows substitution rates from the root of the tree and the node labels indicated bootstrap support values ()
Zervas et al GBE
2900 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
of bchI before the crtB crtI genes in the first cluster For
Methylobacterium spp groups 2 and 3 show similar gene
syntenymdashthough there are a few differences in gene content
(table 2)mdashwhereas groups 1 and 4 appear to also share the
overall architecture of the PGCs For isolates WL9 and WL19
(both in group 1) the exact location of the crtB and crtI genes
is not known as they appear on a short contig by themselves
Annotating the assemblies on RAST revealed several house-
keeping genes present in the vicinity of these clusters which
further supported the notion of their chromosomal place-
ment The completed genomes resulting from the hybrid as-
semblies (strains WL1 WL3 WL4 and WL45) provided the
final evidence of the exact placement of the PGCs on
the chromosomes of all our isolates The different
architectures of the isolated PGCs are shown in figure 5
General descriptive statistics of the assembled complete
genomes are given in table 3 The hybrid assemblies resulted
in both complete genomes and plasmids Thus Rhizobium sp
WL3 contains two plasmids Roseomonas sp WL45 contains
two chromosomes and seven plasmids Alsobacter sp WL4
with two plasmids Methylobacterium sp WL1 with one
plasmid
Using Roseomonas sp WL45 as outgroup we generated
phylogenetic trees for the 16S acsF bchL bchY crtB pufL
and pufM genes as described previously (data not shown)
We calculated substitution rates from the root of the different
trees including the core proteome (CheckM) tree and com-
pared them to each other (table 4)
Discussion
A New Strategy to Generate Complete Genomes ofUnique Aerobic Anoxygenic Phototrophic Bacteria FromEnvironmental Isolates
Aiming to enhance our knowledge of the prevalence of aer-
obic anoxygenic phototrophic bacteria in the phyllosphere
we designed this study using wheat as the plant of choice
and employed a high-throughput multi-disciplinary approach
to identify isolate and analyze AAP strains From the initial
137 plates we ended up with 4480 bacterial colonies
gt500 of which were probable AAP strains After macroscopic
investigation of size shape color and texture of the colonies
we chose 129 strains for which we ran proteome profiling
using MALDI-TOF MS This approach has been shown to be
FIG 3mdashPairwise proteome comparisons (BLAST-matrix) of the 21 bacterial isolates of the analysis The gray-to-green color gradient indicates low to
high percentages of identity The bottom boxes show the presence of homologs within the same proteome (gray-to-red color gradient shows low to high
presence of homologs in the specific proteome)
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2901
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
able to reproducibly distinguish different taxa and in cases
where reference MALDI-TOF spectra are available in
the database be able to assign Genus andor Species names
(Madsen et al 2015) We then proceeded with whole ge-
nome sequencing on the Illumina Nextseq platform and for
the unique genomes we also proceeded with Oxford
Nanopore Technologies sequencing to generate complete
genomes and plasmids
Following this multi-disciplinary approach that relies on
classical microbiological techniques (plating morphology
etc) biochemical analyses (MALDI-TOF MS) and two differ-
ent types of high-throughput whole genome sequencing we
were able to generate draft and complete genomes of novel
AAP bacteria from the wheat phyllosphere in a relatively quick
(weeks) and efficient manner without losing novel biological
information in the process of reducing redundancy
Bacteriochlorophyll-Containing AAP Bacteria in the WheatPhyllosphere
By investigating the presence of bacteriochlorophylls in the
wheat phyllosphere we were able to identify 21 unique strains
harboring them the majority of which belonged to the genus
Methylobacterium Members of this genus are ubiquitous in
nature and have been detected in both soil and plant surfaces
(Green 2006) Targeting the bchY (chlorophyllide reductase
subunit Y) and pufM (reaction centre subunit M) genes in
phyllosphere samples of different plants it was shown that
the Methylobacteriaceae (alpha-Proteobacteria) comprised
28 of the total pufM sequences from the amplicon sequenc-
ing analysis (Atamna-Ismaeel and Finkel 2012a) Our results
are thus in accordance to these previous studies As
Methylobacteriaceae are indigenous inhabitants of the phyllo-
sphere and have consistently been shown to harbor PGCs it
may be that these photosystems are essential to the
Methylobacteria that inhabit this niche Interestingly in their
study Atamna-Ismaeel et al had negative results from their
bchY amplicon sequencing approach suggesting the absence
of bacteria that contain RC1 (bacteriochlorophyll-a containing
reaction centers) type photosystems Our results show that all
isolates contain the complete cassette of chlorophyllide syn-
thase subunit encoding genes (bchB bchC bchF bchG bchI
bchL bchM bchN bchP bchX bchY bchZ) as well as both the
pufL and pufM genes (table 2) Testing the same primers (Yutin
et al 2009) in silico against our bchY sequences we had pos-
itive results for all of them (data not shown)
Patterns of loss of the carotenoid genes have been ob-
served in other alpha and gamma proteobacteria like in
Rhodobacterales and Sphingomonadales (Zheng et al
2011) The absence of crtA results in the inability to perform
the final step in the spheroidene pathway where hydroxy-
spheroidene is converted to hydroxyspheroidenone
Moreover the absence of crtC in Methylobacterium spp of
groups 3 and 4 completely affects both the spheroidene and
the spirilloxanthin pathways as its product is essential for the
first step of both metabolic pathways starting from neuro-
sporene and lycopene respectively This means that strains
FIG 4mdashWhole genome average nucleotide identity (ANI) cladogram
showing percentages of k-mer identities between the isolates Taxa with
the same color and above 90 identity threshold are considered to belong
in the same species Beijerinckia indica ATCC-9039 was used as outgroup
of the analyses (not defined a priori)
Zervas et al GBE
2902 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Tab
le2
Gen
eC
onte
nt
of
the
Photo
synth
esis
Gen
eC
lust
ers
Iden
tified
inth
e21
Bac
terial
Isola
tes
Bact
eri
och
loro
ph
yll
Syn
thase
Gen
es
Caro
ten
oid
Bio
syn
thesi
sG
en
es
Lig
ht-
Harv
est
ing
Co
mp
lex
Gen
es
Str
ain
acs
Fb
chB
bch
Cb
chF
bch
Gb
chI
bch
Lb
chM
bch
Nb
chP
bch
Xb
chY
bch
Zcr
tAcr
tBcr
tCcr
tDcr
tEcr
tFcr
tIcr
tKp
ucC
pu
fAp
ufB
pu
fCp
ufL
pu
fMp
uh
A
Rh
izo
biu
msp
W
L3
Ro
seo
mo
nas
sp
WL4
5
Als
ob
act
er
sp
WL4
Gro
up
1M
eth
ylo
bact
eri
um
sp
WL9
Meth
ylo
bact
eri
um
sp
WL1
9
Meth
ylo
bact
eri
um
sp
WL6
9
Gro
up
2M
eth
ylo
bact
eri
um
sp
WL1
Meth
ylo
bact
eri
um
sp
WL2
Meth
ylo
bact
eri
um
sp
WL7
Meth
ylo
bact
eri
um
sp
WL1
8
Meth
ylo
bact
eri
um
sp
WL6
4
Gro
up
3M
eth
ylo
bact
eri
um
sp
WL6
Meth
ylo
bact
eri
um
sp
WL3
0
Meth
ylo
bact
eri
um
sp
WL9
3
Meth
ylo
bact
eri
um
sp
WL1
16
Meth
ylo
bact
eri
um
sp
WL1
19
Gro
up
4M
eth
ylo
bact
eri
um
sp
WL8
Meth
ylo
bact
eri
um
sp
WL1
2
Meth
ylo
bact
eri
um
sp
WL1
03
Meth
ylo
bact
eri
um
sp
WL1
20
Meth
ylo
bact
eri
um
sp
WL1
22
NO
TEmdash
Gra
yb
oxe
sin
dic
ate
pre
sen
ceo
fg
en
es
an
dw
hit
eb
oxe
sin
dic
ate
ab
sen
ceo
fg
en
es
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2903
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
of these two groups rely on the zeaxanthin pathway to pro-
duce carotenoid pigments and most likely nostoxanthin and
erythroxanthin which however do not participate in the
light-harvesting process but rather have a photoprotection
role (Noguchi et al 1992 Yurkov and Csotonyi 2009)
Phylogenetic Observations
In our study we sequenced and analyzed 21 aerobic anoxy-
genic photosynthetic bacteria most of which belonged to the
genus Methylobacterium the presence of which in the
environment has already been discussed previously
Methylobacterium spp are facultative methylotrophs and
can utilize a variety of C1 C2 C3 and C4 compounds
Such compounds are often found in the phyllosphere due
to plant metabolism (Vorholt 2012 Remus-Emsermann
et al 2014) Methylobacterium spp show a wide range in
chromosome size and number of plasmids they contain
(Marx et al 2012) The genus is represented well in
GenBank with 100 complete or draft genomes (last accessed
April 2019) However most of the sequences do not have
FIG 5mdashArchitecture of the different photosynthetic gene clusters Arrows indicate the orientation of the genes Sizes are not representatives of gene
length Only the genes shared by all PGCs were shown Genes that may be missing from some strains are denoted with parentheses on the figurersquos legend
Table 3
General Descriptive Statistics of the Four Complete AAP Bacterial Genomes
Taxon Genome Size (bp) GC Number of Genes Plasmids Plasmids Size (bp) GC
Rhizobium sp WL3 4568855 6160 4450 1 700631 586
mdash mdash mdash 2 85350 575
Roseomonas sp WL45 4533887 708 5042 1 262815 655
1011854 707 1138 2 240556 662
mdash mdash mdash 3 152922 658
mdash mdash mdash 4 85279 646
mdash mdash mdash 5 84774 661
mdash mdash mdash 6 83273 656
mdash mdash mdash 7 65860 648
Alsobacter sp WL4 5405668 679 5013 1 27178 620
mdash mdash mdash 2 9451 606
Methylobacterium sp WL1 6215463 691 6431 1 35731 638
Zervas et al GBE
2904 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
species affiliation (fig 4) Thus we were not able to assign any
of the 18 isolates to specific Methylobacterium species apart
perhaps from group 2 (WL1 WL2 WL7 WL18 WL64) which
is placed as sister to M brachiatum in the MASH-ANI analysis
Even in this case though M pseudosasicola BL36 is also
placed in the same clade as M brachiatum which raises the
question of proper nomenclature This is further evidenced in
the same tree (cyanide clade below) where seven species
share gt90 of their genome For the remaining isolates
WL19 is closely affiliated to other Methylobacterium strains
that have not been identified A similar picture is shown for
WL69 On the other hand WL19 appears to be a novel spe-
cies with unique genomic features and so do the strains from
groups 3 and 4 (red clade) Thus despite how ubiquitous
Methylobacterium is in nature it is clear that part of its diver-
gence is still missing Moreover a critical evaluation of the
current nomenclature of the genus would be beneficial based
on the whole genome sequences because closely related
genomes (MASH-ANI identitygt 97) have been given differ-
ent species affiliations
Apart from this genus we also identified a Rhizobium sp
a Roseomonas sp and a novel Alsobacter strain (WL4) that
belongs in the Methylocystaceae a family of methanotroph
bacteria This is the first time that AAP bacteria from this
family have been isolated and their complete genomes assem-
bled Its closely related Alsobacter spp have been shown to
exhibit high Thallium tolerance (Bao et al 2006) and it will be
interesting to identify heavy-metal resistance genes in non-
heavy-metal polluted phyllosphere samples It is worth men-
tioning that if WL4 contains heavy-metal resistance genes on
top of the PGC this will have been shown for the first time
and its potential to bioremediate heavy-metal contaminated
areas if applied as plant growth promotion and environmental
remediation agent is of high value and needs to further be
explored
Molecular Evolution of the PGCs
For all gene trees of the PGC we observed that Rhizobium sp
WL3 evolves faster than the other isolates followed by
Table 4
Substitutions Per Site in the Different Phylogenetic Analyses
Strain 16S acsF bchL bchY crtB pufL pufMsuper-matrix
CheckM
Roseomonas sp WL45 0096 0147 0100 0235 0250 0095 0132 0160 0184
Rhizobium sp WL3 0184 0622 0767 0690 1370 0343 0525 0580 0372
Alsobacter sp WL4 0151 0462 0233 0490 0870 0258 0334 0380 0403
Methylobacterium sp WL9 0226 0418 0372 0420 0670 0270 0319 0364 0477
Methylobacterium sp WL19 0221 0413 0300 0460 0600 0233 0346 0364 0468
Methylobacterium sp WL69 0231 0373 0333 0470 0620 0260 0358 0368 0468
Methylobacterium sp WL1 0233 0422 0344 0485 0480 0280 0385 0380 0459
Methylobacterium sp WL2 0233 0422 0344 0485 0480 0280 0385 0380 0459
Methylobacterium sp WL7 0238 0427 0344 0480 0480 0278 0385 0380 0459
Methylobacterium sp WL18 0238 0427 0344 0480 0480 0280 0385 0380 0459
Methylobacterium sp WL64 0233 0422 0344 0480 0485 0280 0385 0380 0459
Methylobacterium sp WL6 0199 0378 0256 0425 0680 0208 0311 0340 0455
Methylobacterium sp WL30 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL93 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL116 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL119 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL8 0188 0373 0258 0430 0680 0230 0358 0348 0459
Methylobacterium sp WL12 0188 0373 0256 0430 0680 0230 0358 0348 0459
Methylobacterium sp WL103 0188 0373 0256 0430 0850 0230 0358 0364 0459
Methylobacterium sp WL120 0188 0373 0258 0430 0680 0233 0358 0348 0459
Methylobacterium sp WL122 0188 0373 0256 0430 0680 0230 0358 0348 0459
Susbstitutions per site
Gro
up1
Gro
up 2
Gro
up 3
Gro
up 4
NOTEmdashAll values correspond to distances from the root where Roseomonas sp WL45 was chosen as outgroup for all the alignments
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2905
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Alsobacter sp WL4 Both these isolates also show much
higher substitution rates compared with their respective 16S
genes and core proteomes (checkM) which indicates that the
PGCs are under relaxed selection pressure In all instances the
phylogenetic trees constructed showed the same topology
and the same grouping of the Methylobacterium isolates
which suggests that the PGCs were incorporated in the
genomes of these strains relatively early in their evolution If
the PGC was free to move laterally then we would have ob-
served very different phylogenetic trees for its genes com-
pared with the 16S and core proteome analyses
Interestingly these observations do not reflect on the archi-
tecture of the PGC where Roseomonas sp WL45 and
Rhizobium sp WL3 contain the PGC in one continuous re-
gion whereas Alsobacter sp WL4 and Methylobacterium sp
WL1 have it in two pieces
For the 18 Methylobacterium isolates group 1 and 2
strains (WL9 WL19 WL69 and WL1 WL2 WL7 WL18
WL64) evolve consistently at different rates compared
with the groups 3 and 4 It appears that there is relatively
strong selection pressure on bchL pufL and pufM which
in all cases are more similar to each other and also show
slower molecular evolution rates compared with ascF
bchY and crtB which in all Methylobacterium isolates
evolve at least twice as fast compared with the 16S
gene (table 4) They also show fewer substitutions per
site compared with the core proteome analysis This result
partially contrasted with previously adopted methods of
using bchY as a marker gene for RC1 clusters (eg
Atamna-Ismaeel and Finkel 2012a) On the other hand
the use of pufM appears more sensible It would how-
ever be more suitable to investigate the possibility of us-
ing other genes for metagenome amplicon sequencing
approaches such as pufL or bchL which show a smaller
degree of divergence in different genera thus universal
primers might be more successful in detecting a wide va-
riety of AAP bacteria in environmental samples
Interestingly while Groups 1 and 2 isolates exhibit higher
substitution rates of the PGC genes compared with
Groups 3 and 4 this is not the case for crtB which evolves
a lot slower and also pufM which shows similar molec-
ular evolution in the genus Overall it seems that the
PGCs were incorporated in ancestral strains that later di-
verged to give rise to the different AAP taxa This is fur-
ther corroborated by the fact that almost all phylogenetic
trees (eg fig 5) are consistent and in agreement with
both the 16S phylogenetic tree (fig 1) as well as the phy-
logenomic tree (fig 2) If the genes of the PGCs had been
transferred to these AAP several times there would have
been significant differences both in the sequence level
and the phylogenetic analysis of the different genes
which would be expected to result in different topologies
and diverse distances from the root even for closelymdash
based on 16S analysismdashrelated strains
The Significance of AAP Bacteria in Wheat Phyllosphere
In this study we isolated for the first time a variety of AAP
bacteria that are present in the wheat phyllosphere adding to
the metagenomics knowledge provided earlier from five other
plants (Atamna-Ismaeel and Finkel 2012a 2012b) The phyllo-
sphere is a rather harsh environment with little water avail-
ability scarce resources very few suitable locations extreme
temperature difference between day and night and high
dosages of UV radiation (Vorholt 2012) AAP bacteria have
the ability to utilize light in photophosphorylation processes
through which they produce ATP and at the same time funnel
the few available nutrients in other metabolic pathways
(Yurkov and Hughes 2017) This has been shown to give
them an edge in vitro after exposure to light compared with
nonAAP bacteria (Ferrera et al 2011) Considering that phyl-
losphere bacteria protect the plant from pathogenic microbes
by secreting metabolites and by occupying the available
niches being able to persevere and outgrow other bacteria
gives a significant edge to AAP taxa and probably points
toward the suggestion of positive selection of such bacteria
in their phyllosphere by the plants themselves as they appear
to be absent from the soil (Atamna-Ismaeel and Finkel
2012b) At this point it is not clear exactly how AAP bacteria
benefit the plant especially since the phyllosphere is a com-
plex habitat with an ever-changing harsh environment In the
past years great efforts have been carried out on how to
improve crop production by studying the rhizosphere It has
also been suggested that the phyllosphere communities play a
significant role in plant health and growth which ultimately
leads to increased yields (Lindow and Brandl 2003) Given the
current projections of world population growth and the sup-
ply of food (UN 2017) it becomes evident that more in-depth
studies investigating the phyllosphere of plants and especially
crops and identifying the underlying drivers of the bacterial
communities and their interactions with their plant hosts is of
paramount importance Thus by isolating maintaining and
analyzing the genomes of phyllosphere isolates we may iden-
tify strains that may prove useful as plant growth promoting
agents in future field trials investigating the effect of phyllo-
sphere inoculation on plant growth health and yield
Supplementary Material
Supplementary data are available at Genome Biology and
Evolution online
Acknowledgments
The authors would like to thank Tue Sparholt Jorgensen for
collecting the wheat sample The authors also want to thank
Tina Thane (AU) Tanja Begovic (AU) and Margit Frederiksen
(NRCWE) for capable laboratory assistance We thank Michal
Koblızek for providing the initial model of the colony infrared
imaging system This work was funded by a Villum
Zervas et al GBE
2906 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Experiment grant (no 17601) and a Marie Skłodowska-Curie
AIAS-COFUND fellowship (EU-FP7 program Grant
Agreement No 609033) awarded to YZ
Author Contributions
AZ YZ LHH designed the study YZ isolated the strains
performed colony infrared imaging and extracted DNA for
Illumina sequencing YZ and AMM performed the MALDI-
TOF MS analysis AZ extracted DNA and performed ONT
Sequencing AZ carried out the bioinformatics and phyloge-
netic analyses AZ and YZ drafted the manuscript AZ
YZ AMM and LHH revised the manuscript for
intellectual content All authors read and approved the
manuscript
Literature Cited
Atamna-Ismaeel N Finkel O 2012a Bacterial anoxygenic photosynthesis
on plant leaf surfaces Environ Microbiol Rep 4(2)209ndash216
Atamna-Ismaeel N Finkel OM 2012b Microbial rhodopsins on leaf sur-
faces of terrestrial plants Environ Microbiol 14(1)140ndash146
Aziz RK et al 2008 The RAST Server rapid annotations using subsystems
technology BMC Genomics 9(1)75
Bankevich A et al 2012 SPAdes a new genome assembly algorithm and
its applications to single-cell sequencing J Comput Biol
19(5)455ndash477
Bao Z Sato Y Kubota M Ohta H 2006 Isolation and characterization of
thallium-tolerant bacteria from heavy metal-polluted river sediment
and non-polluted soils Microb Environ 21(4)251ndash260
Beatie GA 2002 Leaf surface waxes and the process of leaf colonization
by microorganisms In Lindow SE Hecht-Poinar EI Elliott VJ editors
Phyllosphere Microbiology St Paul (USA) APS Press pp 3ndash26
Beattie GA Lindow SE 1999 Bacterial colonization of leaves a spectrum
of strategies Phytopathology 89(5)353ndash359
Brettin T et al 2015 RASTtk a modular and extensible implementation of
the RAST algorithm for building custom annotation pipelines and an-
notating batches of genomes Sci Rep 58365
Carvalho SD Castillo JA 2018 Influence of light on plantndashphyllosphere
interaction Front Plant Sci 91482
Ferrera I Gasol JM Sebastian M Hojerova E Koblızek M 2011
Comparison of growth rates of aerobic anoxygenic phototrophic bac-
teria and other bacterioplankton groups in coastal Mediterranean wa-
ters Appl Environ Microbiol 77(21)7451ndash7458
Ferrera I Sanchez O Kolarova E Koblızek M Gasol JM 2017 Light
enhances the growth rates of natural populations of aerobic anoxy-
genic phototrophic bacteria ISME J 11(10)2391ndash2393
Gorlach K Shingaki R Morisaki H Hattori T 1994 Construction of eco-
collection of paddy field soil bacteria for population analysis J Gen
Appl Microbiol 40(6)509ndash517
Gourion B Rossignol M Vorholt JA Lindow SE 2006 A proteomic study
of Methylobacterium extorquens reveals a response regulator essential
for epiphytic growth wwwintegratedgenomicscom Accessed
November 26 2018
Green PN 2006 Methylobacterium In Dworkin M Falkow S Rosenberg
E Schleifer K Stackebrandt E editors The prokaryotes ndash a handbook
on the biology of bacteria New York Springer p 257ndash265
Hauruseu D Koblızek M 2012 Influence of light on carbon utilization in
aerobic anoxygenic phototrophs Appl Environ Microbiol
78(20)7414ndash7419
Katoh K Misawa K Kuma K Miyata T 2002 MAFFT a novel method for
rapid multiple sequence alignment based on fast Fourier transform
Nucleic Acids Res 30(14)3059ndash3066
Knief C et al 2012 Metaproteogenomic analysis of microbial communi-
ties in the phyllosphere and rhizosphere of rice ISME J
6(7)1378ndash1390
Knief C Frances L Vorholt JA 2010 Competitiveness of diverse methyl-
obacterium strains in the phyllosphere of arabidopsis thaliana and
identification of representative models including M extorquens
PA1 Microb Ecol 60(2)440ndash452
Kwak MJ et al 2014 Genome information of Methylobacterium oryzae a
plantndashprobiotic methylotroph in the phyllosphere PLoS One
9(9)e106704
Lindow SE Brandl MT 2003 Microbiology of the phyllosphere Appl
Environ Microbiol 69(4)1875ndash1883
Madsen AM Zervas A Tendal K Nielsen JL 2015 Microbial diversity in
bioaerosol samples causing ODTS compared to reference bioaerosol
samples as measured using Illumina sequencing and MALDI-TOF
Environ Res 140255ndash267
Martin M 2011 Cutadapt removes adapter sequences from high-
throughput sequencing reads EMBnet J 17(1)10ndash12
Marx CJ et al 2012 Complete genome sequences of six strains of the
genus Methylobacterium b J Bacteriol 194(17)4746
Noguchi T Hayashi H Shimada K Takaichi S Tasumi M 1992 In vivo
states and functions of carotenoids in an aerobic photosynthetic bac-
terium Erythrobacter longus Photosynth Res 31(1)21ndash30
Olm MR Brown CT Brooks B Banfield JF 2017 dRep a tool for fast and
accurate genomic comparisons that enables improved genome recov-
ery from metagenomes through de-replication ISME J
11(12)2864ndash2868
Overbeek R et al 2014 The SEED and the Rapid Annotation of microbial
genomes using Subsystems Technology (RAST) Nucl Acids Res
42(D1)D206ndashD214
Parks DH Imelfort M Skennerton CT Hugenholtz P Tyson GW 2015
CheckM assessing the quality of microbial genomes recovered from
isolates single cells and metagenomes Genome Res 25(7)1043ndash1055
Remus-Emsermann MNP et al 2014 Spatial distribution analyses of nat-
ural phyllosphere-colonizing bacteria on Arabidopsis thaliana revealed
by fluorescence in situ hybridization Environ Microbiol
16(7)2329ndash2340
Stamatakis A 2006 RAxML-VI-HPC maximum likelihood-based phyloge-
netic analyses with thousands of taxa and mixed models
Bioinformatics 22(21)2688ndash2690
UN 2017 World population prospects the 2017 revision Data Booklet
httpspopulationunorgwppPublicationsFilesWPP2019_DataBook-
letpdf
Vesth T Lagesen K Acar euroO Ussery D 2013 CMG-biotools a free work-
bench for basic comparative microbial genomics PLoS One
8(4)e60120
Vorholt JA 2012 Microbial life in the phyllosphere Nat Rev Microbiol
10(12)828ndash840
Wick RR Judd LM Gorrie CL Holt KE 2017 Unicycler resolving bacterial
genome assemblies from short and long sequencing reads PLoS
Comput Biol 13(6)e1005595
Wilson M Lindow SE 1994a Coexistence among epiphytic bacterial pop-
ulations mediated through nutritional resource partitioning Appl
Environ Microbiol 604468ndash77
Wilson M Lindow SE 1994b Ecological similarity and coexistence of epi-
phytic ice-nucleating (ice) pseudomonas syringae strains and a non-
ice-nucleating (ice) biological control agent Appl Environ Microbiol
60(9)3128ndash3137
Woodward FI Lomas MR 2004 Vegetation dynamics ndash simulating
responses to climatic change Biol Rev 79(3)643ndash670
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2907
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Xu J Gordon JI 2003 Honor thy symbionts Proc Natl Acad Sci USA
100(18)10452ndash10459
Yurkov V Csotonyi JT 2009 New light on aerobic anoxygenic photo-
trophs In Hunter NC Daldal F Thurnauer MC Beatty JT editors
The purple photosynthetic bacteria Dordrecht Springer thorn Business
Media BV p 31ndash55 doi101007978-1-4020-8815-5_3
Yurkov V Hughes E 2017 Aerobic Anoxygenic Phototrophs four
decades of mystery In Hallenbeck PC editor Modern topics in the
phototrophic prokaryotes environmental and applied aspects p
193ndash214
Yutin N et al 2009 BchY-based degenerate primers target all types of
anoxygenic photosynthetic bacteria in a single PCR Appl Environ
Microbiol 75(23)7556ndash7559
Zeng Y Feng F Medova H Dean J Koblizek M 2014 Functional
type 2 photosynthetic reaction centers found in the rare bacterial phy-
lum Gemmatimonadetes Proc Natl Acad Sci USA 111(21)7795ndash7800
Zheng Q et al 2011 Diverse arrangement of photosynthetic gene
clusters in aerobic anoxygenic phototrophic bacteria PLoS One
6(9)e25050
Associate editor Esperanza Martinez-Romero
Zervas et al GBE
2908 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
- evz204-TF1
- evz204-TF2
-
Methylobacterium group 3 strains and missing from groups 1
2 and 4 Similar patterns are evidenced in pufB which is ab-
sent from Methylobacterium groups 3 and 4 Rhizobium sp
WL3 and Roseomonas sp WL45 and pufC missing from
Methylobacterium group 4 as well as the novel strain WL4
The remaining genes were found ubiquitously in all strains of
the analysis with only a few exceptions (table 2)
The PGCs were consistently found on the longer contigs of
the initial assemblies of Illumina data produced by SPAdes
(for the strains with fewer than 500 contigs) (table 1) In
Rhizobium sp WL3 and Roseomonas sp WL45 the photosyn-
thetic genes form one continuous cluster with different
however architectures In Alsobacter sp WL4 the genes are
organized in two clusters similar to the Methylobacterium
spp isolates In these cases the genes were found in the
middle of the contigs ruling out the possibility of PGC being
organized into two clusters as an artifact of the assembly
Unique to the architecture present in WL4 is the presence
FIG 2mdashPhylogenomic tree based on the core proteome (CheckM) of the 21 pure bacterial isolates of the analysis including Methylosinus trichosporium
OB3B and Alsobacter sp SH9 The legend shows substitution rates from the root of the tree and the node labels indicated bootstrap support values ()
Zervas et al GBE
2900 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
of bchI before the crtB crtI genes in the first cluster For
Methylobacterium spp groups 2 and 3 show similar gene
syntenymdashthough there are a few differences in gene content
(table 2)mdashwhereas groups 1 and 4 appear to also share the
overall architecture of the PGCs For isolates WL9 and WL19
(both in group 1) the exact location of the crtB and crtI genes
is not known as they appear on a short contig by themselves
Annotating the assemblies on RAST revealed several house-
keeping genes present in the vicinity of these clusters which
further supported the notion of their chromosomal place-
ment The completed genomes resulting from the hybrid as-
semblies (strains WL1 WL3 WL4 and WL45) provided the
final evidence of the exact placement of the PGCs on
the chromosomes of all our isolates The different
architectures of the isolated PGCs are shown in figure 5
General descriptive statistics of the assembled complete
genomes are given in table 3 The hybrid assemblies resulted
in both complete genomes and plasmids Thus Rhizobium sp
WL3 contains two plasmids Roseomonas sp WL45 contains
two chromosomes and seven plasmids Alsobacter sp WL4
with two plasmids Methylobacterium sp WL1 with one
plasmid
Using Roseomonas sp WL45 as outgroup we generated
phylogenetic trees for the 16S acsF bchL bchY crtB pufL
and pufM genes as described previously (data not shown)
We calculated substitution rates from the root of the different
trees including the core proteome (CheckM) tree and com-
pared them to each other (table 4)
Discussion
A New Strategy to Generate Complete Genomes ofUnique Aerobic Anoxygenic Phototrophic Bacteria FromEnvironmental Isolates
Aiming to enhance our knowledge of the prevalence of aer-
obic anoxygenic phototrophic bacteria in the phyllosphere
we designed this study using wheat as the plant of choice
and employed a high-throughput multi-disciplinary approach
to identify isolate and analyze AAP strains From the initial
137 plates we ended up with 4480 bacterial colonies
gt500 of which were probable AAP strains After macroscopic
investigation of size shape color and texture of the colonies
we chose 129 strains for which we ran proteome profiling
using MALDI-TOF MS This approach has been shown to be
FIG 3mdashPairwise proteome comparisons (BLAST-matrix) of the 21 bacterial isolates of the analysis The gray-to-green color gradient indicates low to
high percentages of identity The bottom boxes show the presence of homologs within the same proteome (gray-to-red color gradient shows low to high
presence of homologs in the specific proteome)
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2901
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
able to reproducibly distinguish different taxa and in cases
where reference MALDI-TOF spectra are available in
the database be able to assign Genus andor Species names
(Madsen et al 2015) We then proceeded with whole ge-
nome sequencing on the Illumina Nextseq platform and for
the unique genomes we also proceeded with Oxford
Nanopore Technologies sequencing to generate complete
genomes and plasmids
Following this multi-disciplinary approach that relies on
classical microbiological techniques (plating morphology
etc) biochemical analyses (MALDI-TOF MS) and two differ-
ent types of high-throughput whole genome sequencing we
were able to generate draft and complete genomes of novel
AAP bacteria from the wheat phyllosphere in a relatively quick
(weeks) and efficient manner without losing novel biological
information in the process of reducing redundancy
Bacteriochlorophyll-Containing AAP Bacteria in the WheatPhyllosphere
By investigating the presence of bacteriochlorophylls in the
wheat phyllosphere we were able to identify 21 unique strains
harboring them the majority of which belonged to the genus
Methylobacterium Members of this genus are ubiquitous in
nature and have been detected in both soil and plant surfaces
(Green 2006) Targeting the bchY (chlorophyllide reductase
subunit Y) and pufM (reaction centre subunit M) genes in
phyllosphere samples of different plants it was shown that
the Methylobacteriaceae (alpha-Proteobacteria) comprised
28 of the total pufM sequences from the amplicon sequenc-
ing analysis (Atamna-Ismaeel and Finkel 2012a) Our results
are thus in accordance to these previous studies As
Methylobacteriaceae are indigenous inhabitants of the phyllo-
sphere and have consistently been shown to harbor PGCs it
may be that these photosystems are essential to the
Methylobacteria that inhabit this niche Interestingly in their
study Atamna-Ismaeel et al had negative results from their
bchY amplicon sequencing approach suggesting the absence
of bacteria that contain RC1 (bacteriochlorophyll-a containing
reaction centers) type photosystems Our results show that all
isolates contain the complete cassette of chlorophyllide syn-
thase subunit encoding genes (bchB bchC bchF bchG bchI
bchL bchM bchN bchP bchX bchY bchZ) as well as both the
pufL and pufM genes (table 2) Testing the same primers (Yutin
et al 2009) in silico against our bchY sequences we had pos-
itive results for all of them (data not shown)
Patterns of loss of the carotenoid genes have been ob-
served in other alpha and gamma proteobacteria like in
Rhodobacterales and Sphingomonadales (Zheng et al
2011) The absence of crtA results in the inability to perform
the final step in the spheroidene pathway where hydroxy-
spheroidene is converted to hydroxyspheroidenone
Moreover the absence of crtC in Methylobacterium spp of
groups 3 and 4 completely affects both the spheroidene and
the spirilloxanthin pathways as its product is essential for the
first step of both metabolic pathways starting from neuro-
sporene and lycopene respectively This means that strains
FIG 4mdashWhole genome average nucleotide identity (ANI) cladogram
showing percentages of k-mer identities between the isolates Taxa with
the same color and above 90 identity threshold are considered to belong
in the same species Beijerinckia indica ATCC-9039 was used as outgroup
of the analyses (not defined a priori)
Zervas et al GBE
2902 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Tab
le2
Gen
eC
onte
nt
of
the
Photo
synth
esis
Gen
eC
lust
ers
Iden
tified
inth
e21
Bac
terial
Isola
tes
Bact
eri
och
loro
ph
yll
Syn
thase
Gen
es
Caro
ten
oid
Bio
syn
thesi
sG
en
es
Lig
ht-
Harv
est
ing
Co
mp
lex
Gen
es
Str
ain
acs
Fb
chB
bch
Cb
chF
bch
Gb
chI
bch
Lb
chM
bch
Nb
chP
bch
Xb
chY
bch
Zcr
tAcr
tBcr
tCcr
tDcr
tEcr
tFcr
tIcr
tKp
ucC
pu
fAp
ufB
pu
fCp
ufL
pu
fMp
uh
A
Rh
izo
biu
msp
W
L3
Ro
seo
mo
nas
sp
WL4
5
Als
ob
act
er
sp
WL4
Gro
up
1M
eth
ylo
bact
eri
um
sp
WL9
Meth
ylo
bact
eri
um
sp
WL1
9
Meth
ylo
bact
eri
um
sp
WL6
9
Gro
up
2M
eth
ylo
bact
eri
um
sp
WL1
Meth
ylo
bact
eri
um
sp
WL2
Meth
ylo
bact
eri
um
sp
WL7
Meth
ylo
bact
eri
um
sp
WL1
8
Meth
ylo
bact
eri
um
sp
WL6
4
Gro
up
3M
eth
ylo
bact
eri
um
sp
WL6
Meth
ylo
bact
eri
um
sp
WL3
0
Meth
ylo
bact
eri
um
sp
WL9
3
Meth
ylo
bact
eri
um
sp
WL1
16
Meth
ylo
bact
eri
um
sp
WL1
19
Gro
up
4M
eth
ylo
bact
eri
um
sp
WL8
Meth
ylo
bact
eri
um
sp
WL1
2
Meth
ylo
bact
eri
um
sp
WL1
03
Meth
ylo
bact
eri
um
sp
WL1
20
Meth
ylo
bact
eri
um
sp
WL1
22
NO
TEmdash
Gra
yb
oxe
sin
dic
ate
pre
sen
ceo
fg
en
es
an
dw
hit
eb
oxe
sin
dic
ate
ab
sen
ceo
fg
en
es
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2903
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
of these two groups rely on the zeaxanthin pathway to pro-
duce carotenoid pigments and most likely nostoxanthin and
erythroxanthin which however do not participate in the
light-harvesting process but rather have a photoprotection
role (Noguchi et al 1992 Yurkov and Csotonyi 2009)
Phylogenetic Observations
In our study we sequenced and analyzed 21 aerobic anoxy-
genic photosynthetic bacteria most of which belonged to the
genus Methylobacterium the presence of which in the
environment has already been discussed previously
Methylobacterium spp are facultative methylotrophs and
can utilize a variety of C1 C2 C3 and C4 compounds
Such compounds are often found in the phyllosphere due
to plant metabolism (Vorholt 2012 Remus-Emsermann
et al 2014) Methylobacterium spp show a wide range in
chromosome size and number of plasmids they contain
(Marx et al 2012) The genus is represented well in
GenBank with 100 complete or draft genomes (last accessed
April 2019) However most of the sequences do not have
FIG 5mdashArchitecture of the different photosynthetic gene clusters Arrows indicate the orientation of the genes Sizes are not representatives of gene
length Only the genes shared by all PGCs were shown Genes that may be missing from some strains are denoted with parentheses on the figurersquos legend
Table 3
General Descriptive Statistics of the Four Complete AAP Bacterial Genomes
Taxon Genome Size (bp) GC Number of Genes Plasmids Plasmids Size (bp) GC
Rhizobium sp WL3 4568855 6160 4450 1 700631 586
mdash mdash mdash 2 85350 575
Roseomonas sp WL45 4533887 708 5042 1 262815 655
1011854 707 1138 2 240556 662
mdash mdash mdash 3 152922 658
mdash mdash mdash 4 85279 646
mdash mdash mdash 5 84774 661
mdash mdash mdash 6 83273 656
mdash mdash mdash 7 65860 648
Alsobacter sp WL4 5405668 679 5013 1 27178 620
mdash mdash mdash 2 9451 606
Methylobacterium sp WL1 6215463 691 6431 1 35731 638
Zervas et al GBE
2904 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
species affiliation (fig 4) Thus we were not able to assign any
of the 18 isolates to specific Methylobacterium species apart
perhaps from group 2 (WL1 WL2 WL7 WL18 WL64) which
is placed as sister to M brachiatum in the MASH-ANI analysis
Even in this case though M pseudosasicola BL36 is also
placed in the same clade as M brachiatum which raises the
question of proper nomenclature This is further evidenced in
the same tree (cyanide clade below) where seven species
share gt90 of their genome For the remaining isolates
WL19 is closely affiliated to other Methylobacterium strains
that have not been identified A similar picture is shown for
WL69 On the other hand WL19 appears to be a novel spe-
cies with unique genomic features and so do the strains from
groups 3 and 4 (red clade) Thus despite how ubiquitous
Methylobacterium is in nature it is clear that part of its diver-
gence is still missing Moreover a critical evaluation of the
current nomenclature of the genus would be beneficial based
on the whole genome sequences because closely related
genomes (MASH-ANI identitygt 97) have been given differ-
ent species affiliations
Apart from this genus we also identified a Rhizobium sp
a Roseomonas sp and a novel Alsobacter strain (WL4) that
belongs in the Methylocystaceae a family of methanotroph
bacteria This is the first time that AAP bacteria from this
family have been isolated and their complete genomes assem-
bled Its closely related Alsobacter spp have been shown to
exhibit high Thallium tolerance (Bao et al 2006) and it will be
interesting to identify heavy-metal resistance genes in non-
heavy-metal polluted phyllosphere samples It is worth men-
tioning that if WL4 contains heavy-metal resistance genes on
top of the PGC this will have been shown for the first time
and its potential to bioremediate heavy-metal contaminated
areas if applied as plant growth promotion and environmental
remediation agent is of high value and needs to further be
explored
Molecular Evolution of the PGCs
For all gene trees of the PGC we observed that Rhizobium sp
WL3 evolves faster than the other isolates followed by
Table 4
Substitutions Per Site in the Different Phylogenetic Analyses
Strain 16S acsF bchL bchY crtB pufL pufMsuper-matrix
CheckM
Roseomonas sp WL45 0096 0147 0100 0235 0250 0095 0132 0160 0184
Rhizobium sp WL3 0184 0622 0767 0690 1370 0343 0525 0580 0372
Alsobacter sp WL4 0151 0462 0233 0490 0870 0258 0334 0380 0403
Methylobacterium sp WL9 0226 0418 0372 0420 0670 0270 0319 0364 0477
Methylobacterium sp WL19 0221 0413 0300 0460 0600 0233 0346 0364 0468
Methylobacterium sp WL69 0231 0373 0333 0470 0620 0260 0358 0368 0468
Methylobacterium sp WL1 0233 0422 0344 0485 0480 0280 0385 0380 0459
Methylobacterium sp WL2 0233 0422 0344 0485 0480 0280 0385 0380 0459
Methylobacterium sp WL7 0238 0427 0344 0480 0480 0278 0385 0380 0459
Methylobacterium sp WL18 0238 0427 0344 0480 0480 0280 0385 0380 0459
Methylobacterium sp WL64 0233 0422 0344 0480 0485 0280 0385 0380 0459
Methylobacterium sp WL6 0199 0378 0256 0425 0680 0208 0311 0340 0455
Methylobacterium sp WL30 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL93 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL116 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL119 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL8 0188 0373 0258 0430 0680 0230 0358 0348 0459
Methylobacterium sp WL12 0188 0373 0256 0430 0680 0230 0358 0348 0459
Methylobacterium sp WL103 0188 0373 0256 0430 0850 0230 0358 0364 0459
Methylobacterium sp WL120 0188 0373 0258 0430 0680 0233 0358 0348 0459
Methylobacterium sp WL122 0188 0373 0256 0430 0680 0230 0358 0348 0459
Susbstitutions per site
Gro
up1
Gro
up 2
Gro
up 3
Gro
up 4
NOTEmdashAll values correspond to distances from the root where Roseomonas sp WL45 was chosen as outgroup for all the alignments
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2905
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Alsobacter sp WL4 Both these isolates also show much
higher substitution rates compared with their respective 16S
genes and core proteomes (checkM) which indicates that the
PGCs are under relaxed selection pressure In all instances the
phylogenetic trees constructed showed the same topology
and the same grouping of the Methylobacterium isolates
which suggests that the PGCs were incorporated in the
genomes of these strains relatively early in their evolution If
the PGC was free to move laterally then we would have ob-
served very different phylogenetic trees for its genes com-
pared with the 16S and core proteome analyses
Interestingly these observations do not reflect on the archi-
tecture of the PGC where Roseomonas sp WL45 and
Rhizobium sp WL3 contain the PGC in one continuous re-
gion whereas Alsobacter sp WL4 and Methylobacterium sp
WL1 have it in two pieces
For the 18 Methylobacterium isolates group 1 and 2
strains (WL9 WL19 WL69 and WL1 WL2 WL7 WL18
WL64) evolve consistently at different rates compared
with the groups 3 and 4 It appears that there is relatively
strong selection pressure on bchL pufL and pufM which
in all cases are more similar to each other and also show
slower molecular evolution rates compared with ascF
bchY and crtB which in all Methylobacterium isolates
evolve at least twice as fast compared with the 16S
gene (table 4) They also show fewer substitutions per
site compared with the core proteome analysis This result
partially contrasted with previously adopted methods of
using bchY as a marker gene for RC1 clusters (eg
Atamna-Ismaeel and Finkel 2012a) On the other hand
the use of pufM appears more sensible It would how-
ever be more suitable to investigate the possibility of us-
ing other genes for metagenome amplicon sequencing
approaches such as pufL or bchL which show a smaller
degree of divergence in different genera thus universal
primers might be more successful in detecting a wide va-
riety of AAP bacteria in environmental samples
Interestingly while Groups 1 and 2 isolates exhibit higher
substitution rates of the PGC genes compared with
Groups 3 and 4 this is not the case for crtB which evolves
a lot slower and also pufM which shows similar molec-
ular evolution in the genus Overall it seems that the
PGCs were incorporated in ancestral strains that later di-
verged to give rise to the different AAP taxa This is fur-
ther corroborated by the fact that almost all phylogenetic
trees (eg fig 5) are consistent and in agreement with
both the 16S phylogenetic tree (fig 1) as well as the phy-
logenomic tree (fig 2) If the genes of the PGCs had been
transferred to these AAP several times there would have
been significant differences both in the sequence level
and the phylogenetic analysis of the different genes
which would be expected to result in different topologies
and diverse distances from the root even for closelymdash
based on 16S analysismdashrelated strains
The Significance of AAP Bacteria in Wheat Phyllosphere
In this study we isolated for the first time a variety of AAP
bacteria that are present in the wheat phyllosphere adding to
the metagenomics knowledge provided earlier from five other
plants (Atamna-Ismaeel and Finkel 2012a 2012b) The phyllo-
sphere is a rather harsh environment with little water avail-
ability scarce resources very few suitable locations extreme
temperature difference between day and night and high
dosages of UV radiation (Vorholt 2012) AAP bacteria have
the ability to utilize light in photophosphorylation processes
through which they produce ATP and at the same time funnel
the few available nutrients in other metabolic pathways
(Yurkov and Hughes 2017) This has been shown to give
them an edge in vitro after exposure to light compared with
nonAAP bacteria (Ferrera et al 2011) Considering that phyl-
losphere bacteria protect the plant from pathogenic microbes
by secreting metabolites and by occupying the available
niches being able to persevere and outgrow other bacteria
gives a significant edge to AAP taxa and probably points
toward the suggestion of positive selection of such bacteria
in their phyllosphere by the plants themselves as they appear
to be absent from the soil (Atamna-Ismaeel and Finkel
2012b) At this point it is not clear exactly how AAP bacteria
benefit the plant especially since the phyllosphere is a com-
plex habitat with an ever-changing harsh environment In the
past years great efforts have been carried out on how to
improve crop production by studying the rhizosphere It has
also been suggested that the phyllosphere communities play a
significant role in plant health and growth which ultimately
leads to increased yields (Lindow and Brandl 2003) Given the
current projections of world population growth and the sup-
ply of food (UN 2017) it becomes evident that more in-depth
studies investigating the phyllosphere of plants and especially
crops and identifying the underlying drivers of the bacterial
communities and their interactions with their plant hosts is of
paramount importance Thus by isolating maintaining and
analyzing the genomes of phyllosphere isolates we may iden-
tify strains that may prove useful as plant growth promoting
agents in future field trials investigating the effect of phyllo-
sphere inoculation on plant growth health and yield
Supplementary Material
Supplementary data are available at Genome Biology and
Evolution online
Acknowledgments
The authors would like to thank Tue Sparholt Jorgensen for
collecting the wheat sample The authors also want to thank
Tina Thane (AU) Tanja Begovic (AU) and Margit Frederiksen
(NRCWE) for capable laboratory assistance We thank Michal
Koblızek for providing the initial model of the colony infrared
imaging system This work was funded by a Villum
Zervas et al GBE
2906 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Experiment grant (no 17601) and a Marie Skłodowska-Curie
AIAS-COFUND fellowship (EU-FP7 program Grant
Agreement No 609033) awarded to YZ
Author Contributions
AZ YZ LHH designed the study YZ isolated the strains
performed colony infrared imaging and extracted DNA for
Illumina sequencing YZ and AMM performed the MALDI-
TOF MS analysis AZ extracted DNA and performed ONT
Sequencing AZ carried out the bioinformatics and phyloge-
netic analyses AZ and YZ drafted the manuscript AZ
YZ AMM and LHH revised the manuscript for
intellectual content All authors read and approved the
manuscript
Literature Cited
Atamna-Ismaeel N Finkel O 2012a Bacterial anoxygenic photosynthesis
on plant leaf surfaces Environ Microbiol Rep 4(2)209ndash216
Atamna-Ismaeel N Finkel OM 2012b Microbial rhodopsins on leaf sur-
faces of terrestrial plants Environ Microbiol 14(1)140ndash146
Aziz RK et al 2008 The RAST Server rapid annotations using subsystems
technology BMC Genomics 9(1)75
Bankevich A et al 2012 SPAdes a new genome assembly algorithm and
its applications to single-cell sequencing J Comput Biol
19(5)455ndash477
Bao Z Sato Y Kubota M Ohta H 2006 Isolation and characterization of
thallium-tolerant bacteria from heavy metal-polluted river sediment
and non-polluted soils Microb Environ 21(4)251ndash260
Beatie GA 2002 Leaf surface waxes and the process of leaf colonization
by microorganisms In Lindow SE Hecht-Poinar EI Elliott VJ editors
Phyllosphere Microbiology St Paul (USA) APS Press pp 3ndash26
Beattie GA Lindow SE 1999 Bacterial colonization of leaves a spectrum
of strategies Phytopathology 89(5)353ndash359
Brettin T et al 2015 RASTtk a modular and extensible implementation of
the RAST algorithm for building custom annotation pipelines and an-
notating batches of genomes Sci Rep 58365
Carvalho SD Castillo JA 2018 Influence of light on plantndashphyllosphere
interaction Front Plant Sci 91482
Ferrera I Gasol JM Sebastian M Hojerova E Koblızek M 2011
Comparison of growth rates of aerobic anoxygenic phototrophic bac-
teria and other bacterioplankton groups in coastal Mediterranean wa-
ters Appl Environ Microbiol 77(21)7451ndash7458
Ferrera I Sanchez O Kolarova E Koblızek M Gasol JM 2017 Light
enhances the growth rates of natural populations of aerobic anoxy-
genic phototrophic bacteria ISME J 11(10)2391ndash2393
Gorlach K Shingaki R Morisaki H Hattori T 1994 Construction of eco-
collection of paddy field soil bacteria for population analysis J Gen
Appl Microbiol 40(6)509ndash517
Gourion B Rossignol M Vorholt JA Lindow SE 2006 A proteomic study
of Methylobacterium extorquens reveals a response regulator essential
for epiphytic growth wwwintegratedgenomicscom Accessed
November 26 2018
Green PN 2006 Methylobacterium In Dworkin M Falkow S Rosenberg
E Schleifer K Stackebrandt E editors The prokaryotes ndash a handbook
on the biology of bacteria New York Springer p 257ndash265
Hauruseu D Koblızek M 2012 Influence of light on carbon utilization in
aerobic anoxygenic phototrophs Appl Environ Microbiol
78(20)7414ndash7419
Katoh K Misawa K Kuma K Miyata T 2002 MAFFT a novel method for
rapid multiple sequence alignment based on fast Fourier transform
Nucleic Acids Res 30(14)3059ndash3066
Knief C et al 2012 Metaproteogenomic analysis of microbial communi-
ties in the phyllosphere and rhizosphere of rice ISME J
6(7)1378ndash1390
Knief C Frances L Vorholt JA 2010 Competitiveness of diverse methyl-
obacterium strains in the phyllosphere of arabidopsis thaliana and
identification of representative models including M extorquens
PA1 Microb Ecol 60(2)440ndash452
Kwak MJ et al 2014 Genome information of Methylobacterium oryzae a
plantndashprobiotic methylotroph in the phyllosphere PLoS One
9(9)e106704
Lindow SE Brandl MT 2003 Microbiology of the phyllosphere Appl
Environ Microbiol 69(4)1875ndash1883
Madsen AM Zervas A Tendal K Nielsen JL 2015 Microbial diversity in
bioaerosol samples causing ODTS compared to reference bioaerosol
samples as measured using Illumina sequencing and MALDI-TOF
Environ Res 140255ndash267
Martin M 2011 Cutadapt removes adapter sequences from high-
throughput sequencing reads EMBnet J 17(1)10ndash12
Marx CJ et al 2012 Complete genome sequences of six strains of the
genus Methylobacterium b J Bacteriol 194(17)4746
Noguchi T Hayashi H Shimada K Takaichi S Tasumi M 1992 In vivo
states and functions of carotenoids in an aerobic photosynthetic bac-
terium Erythrobacter longus Photosynth Res 31(1)21ndash30
Olm MR Brown CT Brooks B Banfield JF 2017 dRep a tool for fast and
accurate genomic comparisons that enables improved genome recov-
ery from metagenomes through de-replication ISME J
11(12)2864ndash2868
Overbeek R et al 2014 The SEED and the Rapid Annotation of microbial
genomes using Subsystems Technology (RAST) Nucl Acids Res
42(D1)D206ndashD214
Parks DH Imelfort M Skennerton CT Hugenholtz P Tyson GW 2015
CheckM assessing the quality of microbial genomes recovered from
isolates single cells and metagenomes Genome Res 25(7)1043ndash1055
Remus-Emsermann MNP et al 2014 Spatial distribution analyses of nat-
ural phyllosphere-colonizing bacteria on Arabidopsis thaliana revealed
by fluorescence in situ hybridization Environ Microbiol
16(7)2329ndash2340
Stamatakis A 2006 RAxML-VI-HPC maximum likelihood-based phyloge-
netic analyses with thousands of taxa and mixed models
Bioinformatics 22(21)2688ndash2690
UN 2017 World population prospects the 2017 revision Data Booklet
httpspopulationunorgwppPublicationsFilesWPP2019_DataBook-
letpdf
Vesth T Lagesen K Acar euroO Ussery D 2013 CMG-biotools a free work-
bench for basic comparative microbial genomics PLoS One
8(4)e60120
Vorholt JA 2012 Microbial life in the phyllosphere Nat Rev Microbiol
10(12)828ndash840
Wick RR Judd LM Gorrie CL Holt KE 2017 Unicycler resolving bacterial
genome assemblies from short and long sequencing reads PLoS
Comput Biol 13(6)e1005595
Wilson M Lindow SE 1994a Coexistence among epiphytic bacterial pop-
ulations mediated through nutritional resource partitioning Appl
Environ Microbiol 604468ndash77
Wilson M Lindow SE 1994b Ecological similarity and coexistence of epi-
phytic ice-nucleating (ice) pseudomonas syringae strains and a non-
ice-nucleating (ice) biological control agent Appl Environ Microbiol
60(9)3128ndash3137
Woodward FI Lomas MR 2004 Vegetation dynamics ndash simulating
responses to climatic change Biol Rev 79(3)643ndash670
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2907
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Xu J Gordon JI 2003 Honor thy symbionts Proc Natl Acad Sci USA
100(18)10452ndash10459
Yurkov V Csotonyi JT 2009 New light on aerobic anoxygenic photo-
trophs In Hunter NC Daldal F Thurnauer MC Beatty JT editors
The purple photosynthetic bacteria Dordrecht Springer thorn Business
Media BV p 31ndash55 doi101007978-1-4020-8815-5_3
Yurkov V Hughes E 2017 Aerobic Anoxygenic Phototrophs four
decades of mystery In Hallenbeck PC editor Modern topics in the
phototrophic prokaryotes environmental and applied aspects p
193ndash214
Yutin N et al 2009 BchY-based degenerate primers target all types of
anoxygenic photosynthetic bacteria in a single PCR Appl Environ
Microbiol 75(23)7556ndash7559
Zeng Y Feng F Medova H Dean J Koblizek M 2014 Functional
type 2 photosynthetic reaction centers found in the rare bacterial phy-
lum Gemmatimonadetes Proc Natl Acad Sci USA 111(21)7795ndash7800
Zheng Q et al 2011 Diverse arrangement of photosynthetic gene
clusters in aerobic anoxygenic phototrophic bacteria PLoS One
6(9)e25050
Associate editor Esperanza Martinez-Romero
Zervas et al GBE
2908 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
- evz204-TF1
- evz204-TF2
-
of bchI before the crtB crtI genes in the first cluster For
Methylobacterium spp groups 2 and 3 show similar gene
syntenymdashthough there are a few differences in gene content
(table 2)mdashwhereas groups 1 and 4 appear to also share the
overall architecture of the PGCs For isolates WL9 and WL19
(both in group 1) the exact location of the crtB and crtI genes
is not known as they appear on a short contig by themselves
Annotating the assemblies on RAST revealed several house-
keeping genes present in the vicinity of these clusters which
further supported the notion of their chromosomal place-
ment The completed genomes resulting from the hybrid as-
semblies (strains WL1 WL3 WL4 and WL45) provided the
final evidence of the exact placement of the PGCs on
the chromosomes of all our isolates The different
architectures of the isolated PGCs are shown in figure 5
General descriptive statistics of the assembled complete
genomes are given in table 3 The hybrid assemblies resulted
in both complete genomes and plasmids Thus Rhizobium sp
WL3 contains two plasmids Roseomonas sp WL45 contains
two chromosomes and seven plasmids Alsobacter sp WL4
with two plasmids Methylobacterium sp WL1 with one
plasmid
Using Roseomonas sp WL45 as outgroup we generated
phylogenetic trees for the 16S acsF bchL bchY crtB pufL
and pufM genes as described previously (data not shown)
We calculated substitution rates from the root of the different
trees including the core proteome (CheckM) tree and com-
pared them to each other (table 4)
Discussion
A New Strategy to Generate Complete Genomes ofUnique Aerobic Anoxygenic Phototrophic Bacteria FromEnvironmental Isolates
Aiming to enhance our knowledge of the prevalence of aer-
obic anoxygenic phototrophic bacteria in the phyllosphere
we designed this study using wheat as the plant of choice
and employed a high-throughput multi-disciplinary approach
to identify isolate and analyze AAP strains From the initial
137 plates we ended up with 4480 bacterial colonies
gt500 of which were probable AAP strains After macroscopic
investigation of size shape color and texture of the colonies
we chose 129 strains for which we ran proteome profiling
using MALDI-TOF MS This approach has been shown to be
FIG 3mdashPairwise proteome comparisons (BLAST-matrix) of the 21 bacterial isolates of the analysis The gray-to-green color gradient indicates low to
high percentages of identity The bottom boxes show the presence of homologs within the same proteome (gray-to-red color gradient shows low to high
presence of homologs in the specific proteome)
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2901
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
able to reproducibly distinguish different taxa and in cases
where reference MALDI-TOF spectra are available in
the database be able to assign Genus andor Species names
(Madsen et al 2015) We then proceeded with whole ge-
nome sequencing on the Illumina Nextseq platform and for
the unique genomes we also proceeded with Oxford
Nanopore Technologies sequencing to generate complete
genomes and plasmids
Following this multi-disciplinary approach that relies on
classical microbiological techniques (plating morphology
etc) biochemical analyses (MALDI-TOF MS) and two differ-
ent types of high-throughput whole genome sequencing we
were able to generate draft and complete genomes of novel
AAP bacteria from the wheat phyllosphere in a relatively quick
(weeks) and efficient manner without losing novel biological
information in the process of reducing redundancy
Bacteriochlorophyll-Containing AAP Bacteria in the WheatPhyllosphere
By investigating the presence of bacteriochlorophylls in the
wheat phyllosphere we were able to identify 21 unique strains
harboring them the majority of which belonged to the genus
Methylobacterium Members of this genus are ubiquitous in
nature and have been detected in both soil and plant surfaces
(Green 2006) Targeting the bchY (chlorophyllide reductase
subunit Y) and pufM (reaction centre subunit M) genes in
phyllosphere samples of different plants it was shown that
the Methylobacteriaceae (alpha-Proteobacteria) comprised
28 of the total pufM sequences from the amplicon sequenc-
ing analysis (Atamna-Ismaeel and Finkel 2012a) Our results
are thus in accordance to these previous studies As
Methylobacteriaceae are indigenous inhabitants of the phyllo-
sphere and have consistently been shown to harbor PGCs it
may be that these photosystems are essential to the
Methylobacteria that inhabit this niche Interestingly in their
study Atamna-Ismaeel et al had negative results from their
bchY amplicon sequencing approach suggesting the absence
of bacteria that contain RC1 (bacteriochlorophyll-a containing
reaction centers) type photosystems Our results show that all
isolates contain the complete cassette of chlorophyllide syn-
thase subunit encoding genes (bchB bchC bchF bchG bchI
bchL bchM bchN bchP bchX bchY bchZ) as well as both the
pufL and pufM genes (table 2) Testing the same primers (Yutin
et al 2009) in silico against our bchY sequences we had pos-
itive results for all of them (data not shown)
Patterns of loss of the carotenoid genes have been ob-
served in other alpha and gamma proteobacteria like in
Rhodobacterales and Sphingomonadales (Zheng et al
2011) The absence of crtA results in the inability to perform
the final step in the spheroidene pathway where hydroxy-
spheroidene is converted to hydroxyspheroidenone
Moreover the absence of crtC in Methylobacterium spp of
groups 3 and 4 completely affects both the spheroidene and
the spirilloxanthin pathways as its product is essential for the
first step of both metabolic pathways starting from neuro-
sporene and lycopene respectively This means that strains
FIG 4mdashWhole genome average nucleotide identity (ANI) cladogram
showing percentages of k-mer identities between the isolates Taxa with
the same color and above 90 identity threshold are considered to belong
in the same species Beijerinckia indica ATCC-9039 was used as outgroup
of the analyses (not defined a priori)
Zervas et al GBE
2902 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Tab
le2
Gen
eC
onte
nt
of
the
Photo
synth
esis
Gen
eC
lust
ers
Iden
tified
inth
e21
Bac
terial
Isola
tes
Bact
eri
och
loro
ph
yll
Syn
thase
Gen
es
Caro
ten
oid
Bio
syn
thesi
sG
en
es
Lig
ht-
Harv
est
ing
Co
mp
lex
Gen
es
Str
ain
acs
Fb
chB
bch
Cb
chF
bch
Gb
chI
bch
Lb
chM
bch
Nb
chP
bch
Xb
chY
bch
Zcr
tAcr
tBcr
tCcr
tDcr
tEcr
tFcr
tIcr
tKp
ucC
pu
fAp
ufB
pu
fCp
ufL
pu
fMp
uh
A
Rh
izo
biu
msp
W
L3
Ro
seo
mo
nas
sp
WL4
5
Als
ob
act
er
sp
WL4
Gro
up
1M
eth
ylo
bact
eri
um
sp
WL9
Meth
ylo
bact
eri
um
sp
WL1
9
Meth
ylo
bact
eri
um
sp
WL6
9
Gro
up
2M
eth
ylo
bact
eri
um
sp
WL1
Meth
ylo
bact
eri
um
sp
WL2
Meth
ylo
bact
eri
um
sp
WL7
Meth
ylo
bact
eri
um
sp
WL1
8
Meth
ylo
bact
eri
um
sp
WL6
4
Gro
up
3M
eth
ylo
bact
eri
um
sp
WL6
Meth
ylo
bact
eri
um
sp
WL3
0
Meth
ylo
bact
eri
um
sp
WL9
3
Meth
ylo
bact
eri
um
sp
WL1
16
Meth
ylo
bact
eri
um
sp
WL1
19
Gro
up
4M
eth
ylo
bact
eri
um
sp
WL8
Meth
ylo
bact
eri
um
sp
WL1
2
Meth
ylo
bact
eri
um
sp
WL1
03
Meth
ylo
bact
eri
um
sp
WL1
20
Meth
ylo
bact
eri
um
sp
WL1
22
NO
TEmdash
Gra
yb
oxe
sin
dic
ate
pre
sen
ceo
fg
en
es
an
dw
hit
eb
oxe
sin
dic
ate
ab
sen
ceo
fg
en
es
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2903
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
of these two groups rely on the zeaxanthin pathway to pro-
duce carotenoid pigments and most likely nostoxanthin and
erythroxanthin which however do not participate in the
light-harvesting process but rather have a photoprotection
role (Noguchi et al 1992 Yurkov and Csotonyi 2009)
Phylogenetic Observations
In our study we sequenced and analyzed 21 aerobic anoxy-
genic photosynthetic bacteria most of which belonged to the
genus Methylobacterium the presence of which in the
environment has already been discussed previously
Methylobacterium spp are facultative methylotrophs and
can utilize a variety of C1 C2 C3 and C4 compounds
Such compounds are often found in the phyllosphere due
to plant metabolism (Vorholt 2012 Remus-Emsermann
et al 2014) Methylobacterium spp show a wide range in
chromosome size and number of plasmids they contain
(Marx et al 2012) The genus is represented well in
GenBank with 100 complete or draft genomes (last accessed
April 2019) However most of the sequences do not have
FIG 5mdashArchitecture of the different photosynthetic gene clusters Arrows indicate the orientation of the genes Sizes are not representatives of gene
length Only the genes shared by all PGCs were shown Genes that may be missing from some strains are denoted with parentheses on the figurersquos legend
Table 3
General Descriptive Statistics of the Four Complete AAP Bacterial Genomes
Taxon Genome Size (bp) GC Number of Genes Plasmids Plasmids Size (bp) GC
Rhizobium sp WL3 4568855 6160 4450 1 700631 586
mdash mdash mdash 2 85350 575
Roseomonas sp WL45 4533887 708 5042 1 262815 655
1011854 707 1138 2 240556 662
mdash mdash mdash 3 152922 658
mdash mdash mdash 4 85279 646
mdash mdash mdash 5 84774 661
mdash mdash mdash 6 83273 656
mdash mdash mdash 7 65860 648
Alsobacter sp WL4 5405668 679 5013 1 27178 620
mdash mdash mdash 2 9451 606
Methylobacterium sp WL1 6215463 691 6431 1 35731 638
Zervas et al GBE
2904 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
species affiliation (fig 4) Thus we were not able to assign any
of the 18 isolates to specific Methylobacterium species apart
perhaps from group 2 (WL1 WL2 WL7 WL18 WL64) which
is placed as sister to M brachiatum in the MASH-ANI analysis
Even in this case though M pseudosasicola BL36 is also
placed in the same clade as M brachiatum which raises the
question of proper nomenclature This is further evidenced in
the same tree (cyanide clade below) where seven species
share gt90 of their genome For the remaining isolates
WL19 is closely affiliated to other Methylobacterium strains
that have not been identified A similar picture is shown for
WL69 On the other hand WL19 appears to be a novel spe-
cies with unique genomic features and so do the strains from
groups 3 and 4 (red clade) Thus despite how ubiquitous
Methylobacterium is in nature it is clear that part of its diver-
gence is still missing Moreover a critical evaluation of the
current nomenclature of the genus would be beneficial based
on the whole genome sequences because closely related
genomes (MASH-ANI identitygt 97) have been given differ-
ent species affiliations
Apart from this genus we also identified a Rhizobium sp
a Roseomonas sp and a novel Alsobacter strain (WL4) that
belongs in the Methylocystaceae a family of methanotroph
bacteria This is the first time that AAP bacteria from this
family have been isolated and their complete genomes assem-
bled Its closely related Alsobacter spp have been shown to
exhibit high Thallium tolerance (Bao et al 2006) and it will be
interesting to identify heavy-metal resistance genes in non-
heavy-metal polluted phyllosphere samples It is worth men-
tioning that if WL4 contains heavy-metal resistance genes on
top of the PGC this will have been shown for the first time
and its potential to bioremediate heavy-metal contaminated
areas if applied as plant growth promotion and environmental
remediation agent is of high value and needs to further be
explored
Molecular Evolution of the PGCs
For all gene trees of the PGC we observed that Rhizobium sp
WL3 evolves faster than the other isolates followed by
Table 4
Substitutions Per Site in the Different Phylogenetic Analyses
Strain 16S acsF bchL bchY crtB pufL pufMsuper-matrix
CheckM
Roseomonas sp WL45 0096 0147 0100 0235 0250 0095 0132 0160 0184
Rhizobium sp WL3 0184 0622 0767 0690 1370 0343 0525 0580 0372
Alsobacter sp WL4 0151 0462 0233 0490 0870 0258 0334 0380 0403
Methylobacterium sp WL9 0226 0418 0372 0420 0670 0270 0319 0364 0477
Methylobacterium sp WL19 0221 0413 0300 0460 0600 0233 0346 0364 0468
Methylobacterium sp WL69 0231 0373 0333 0470 0620 0260 0358 0368 0468
Methylobacterium sp WL1 0233 0422 0344 0485 0480 0280 0385 0380 0459
Methylobacterium sp WL2 0233 0422 0344 0485 0480 0280 0385 0380 0459
Methylobacterium sp WL7 0238 0427 0344 0480 0480 0278 0385 0380 0459
Methylobacterium sp WL18 0238 0427 0344 0480 0480 0280 0385 0380 0459
Methylobacterium sp WL64 0233 0422 0344 0480 0485 0280 0385 0380 0459
Methylobacterium sp WL6 0199 0378 0256 0425 0680 0208 0311 0340 0455
Methylobacterium sp WL30 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL93 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL116 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL119 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL8 0188 0373 0258 0430 0680 0230 0358 0348 0459
Methylobacterium sp WL12 0188 0373 0256 0430 0680 0230 0358 0348 0459
Methylobacterium sp WL103 0188 0373 0256 0430 0850 0230 0358 0364 0459
Methylobacterium sp WL120 0188 0373 0258 0430 0680 0233 0358 0348 0459
Methylobacterium sp WL122 0188 0373 0256 0430 0680 0230 0358 0348 0459
Susbstitutions per site
Gro
up1
Gro
up 2
Gro
up 3
Gro
up 4
NOTEmdashAll values correspond to distances from the root where Roseomonas sp WL45 was chosen as outgroup for all the alignments
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2905
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Alsobacter sp WL4 Both these isolates also show much
higher substitution rates compared with their respective 16S
genes and core proteomes (checkM) which indicates that the
PGCs are under relaxed selection pressure In all instances the
phylogenetic trees constructed showed the same topology
and the same grouping of the Methylobacterium isolates
which suggests that the PGCs were incorporated in the
genomes of these strains relatively early in their evolution If
the PGC was free to move laterally then we would have ob-
served very different phylogenetic trees for its genes com-
pared with the 16S and core proteome analyses
Interestingly these observations do not reflect on the archi-
tecture of the PGC where Roseomonas sp WL45 and
Rhizobium sp WL3 contain the PGC in one continuous re-
gion whereas Alsobacter sp WL4 and Methylobacterium sp
WL1 have it in two pieces
For the 18 Methylobacterium isolates group 1 and 2
strains (WL9 WL19 WL69 and WL1 WL2 WL7 WL18
WL64) evolve consistently at different rates compared
with the groups 3 and 4 It appears that there is relatively
strong selection pressure on bchL pufL and pufM which
in all cases are more similar to each other and also show
slower molecular evolution rates compared with ascF
bchY and crtB which in all Methylobacterium isolates
evolve at least twice as fast compared with the 16S
gene (table 4) They also show fewer substitutions per
site compared with the core proteome analysis This result
partially contrasted with previously adopted methods of
using bchY as a marker gene for RC1 clusters (eg
Atamna-Ismaeel and Finkel 2012a) On the other hand
the use of pufM appears more sensible It would how-
ever be more suitable to investigate the possibility of us-
ing other genes for metagenome amplicon sequencing
approaches such as pufL or bchL which show a smaller
degree of divergence in different genera thus universal
primers might be more successful in detecting a wide va-
riety of AAP bacteria in environmental samples
Interestingly while Groups 1 and 2 isolates exhibit higher
substitution rates of the PGC genes compared with
Groups 3 and 4 this is not the case for crtB which evolves
a lot slower and also pufM which shows similar molec-
ular evolution in the genus Overall it seems that the
PGCs were incorporated in ancestral strains that later di-
verged to give rise to the different AAP taxa This is fur-
ther corroborated by the fact that almost all phylogenetic
trees (eg fig 5) are consistent and in agreement with
both the 16S phylogenetic tree (fig 1) as well as the phy-
logenomic tree (fig 2) If the genes of the PGCs had been
transferred to these AAP several times there would have
been significant differences both in the sequence level
and the phylogenetic analysis of the different genes
which would be expected to result in different topologies
and diverse distances from the root even for closelymdash
based on 16S analysismdashrelated strains
The Significance of AAP Bacteria in Wheat Phyllosphere
In this study we isolated for the first time a variety of AAP
bacteria that are present in the wheat phyllosphere adding to
the metagenomics knowledge provided earlier from five other
plants (Atamna-Ismaeel and Finkel 2012a 2012b) The phyllo-
sphere is a rather harsh environment with little water avail-
ability scarce resources very few suitable locations extreme
temperature difference between day and night and high
dosages of UV radiation (Vorholt 2012) AAP bacteria have
the ability to utilize light in photophosphorylation processes
through which they produce ATP and at the same time funnel
the few available nutrients in other metabolic pathways
(Yurkov and Hughes 2017) This has been shown to give
them an edge in vitro after exposure to light compared with
nonAAP bacteria (Ferrera et al 2011) Considering that phyl-
losphere bacteria protect the plant from pathogenic microbes
by secreting metabolites and by occupying the available
niches being able to persevere and outgrow other bacteria
gives a significant edge to AAP taxa and probably points
toward the suggestion of positive selection of such bacteria
in their phyllosphere by the plants themselves as they appear
to be absent from the soil (Atamna-Ismaeel and Finkel
2012b) At this point it is not clear exactly how AAP bacteria
benefit the plant especially since the phyllosphere is a com-
plex habitat with an ever-changing harsh environment In the
past years great efforts have been carried out on how to
improve crop production by studying the rhizosphere It has
also been suggested that the phyllosphere communities play a
significant role in plant health and growth which ultimately
leads to increased yields (Lindow and Brandl 2003) Given the
current projections of world population growth and the sup-
ply of food (UN 2017) it becomes evident that more in-depth
studies investigating the phyllosphere of plants and especially
crops and identifying the underlying drivers of the bacterial
communities and their interactions with their plant hosts is of
paramount importance Thus by isolating maintaining and
analyzing the genomes of phyllosphere isolates we may iden-
tify strains that may prove useful as plant growth promoting
agents in future field trials investigating the effect of phyllo-
sphere inoculation on plant growth health and yield
Supplementary Material
Supplementary data are available at Genome Biology and
Evolution online
Acknowledgments
The authors would like to thank Tue Sparholt Jorgensen for
collecting the wheat sample The authors also want to thank
Tina Thane (AU) Tanja Begovic (AU) and Margit Frederiksen
(NRCWE) for capable laboratory assistance We thank Michal
Koblızek for providing the initial model of the colony infrared
imaging system This work was funded by a Villum
Zervas et al GBE
2906 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Experiment grant (no 17601) and a Marie Skłodowska-Curie
AIAS-COFUND fellowship (EU-FP7 program Grant
Agreement No 609033) awarded to YZ
Author Contributions
AZ YZ LHH designed the study YZ isolated the strains
performed colony infrared imaging and extracted DNA for
Illumina sequencing YZ and AMM performed the MALDI-
TOF MS analysis AZ extracted DNA and performed ONT
Sequencing AZ carried out the bioinformatics and phyloge-
netic analyses AZ and YZ drafted the manuscript AZ
YZ AMM and LHH revised the manuscript for
intellectual content All authors read and approved the
manuscript
Literature Cited
Atamna-Ismaeel N Finkel O 2012a Bacterial anoxygenic photosynthesis
on plant leaf surfaces Environ Microbiol Rep 4(2)209ndash216
Atamna-Ismaeel N Finkel OM 2012b Microbial rhodopsins on leaf sur-
faces of terrestrial plants Environ Microbiol 14(1)140ndash146
Aziz RK et al 2008 The RAST Server rapid annotations using subsystems
technology BMC Genomics 9(1)75
Bankevich A et al 2012 SPAdes a new genome assembly algorithm and
its applications to single-cell sequencing J Comput Biol
19(5)455ndash477
Bao Z Sato Y Kubota M Ohta H 2006 Isolation and characterization of
thallium-tolerant bacteria from heavy metal-polluted river sediment
and non-polluted soils Microb Environ 21(4)251ndash260
Beatie GA 2002 Leaf surface waxes and the process of leaf colonization
by microorganisms In Lindow SE Hecht-Poinar EI Elliott VJ editors
Phyllosphere Microbiology St Paul (USA) APS Press pp 3ndash26
Beattie GA Lindow SE 1999 Bacterial colonization of leaves a spectrum
of strategies Phytopathology 89(5)353ndash359
Brettin T et al 2015 RASTtk a modular and extensible implementation of
the RAST algorithm for building custom annotation pipelines and an-
notating batches of genomes Sci Rep 58365
Carvalho SD Castillo JA 2018 Influence of light on plantndashphyllosphere
interaction Front Plant Sci 91482
Ferrera I Gasol JM Sebastian M Hojerova E Koblızek M 2011
Comparison of growth rates of aerobic anoxygenic phototrophic bac-
teria and other bacterioplankton groups in coastal Mediterranean wa-
ters Appl Environ Microbiol 77(21)7451ndash7458
Ferrera I Sanchez O Kolarova E Koblızek M Gasol JM 2017 Light
enhances the growth rates of natural populations of aerobic anoxy-
genic phototrophic bacteria ISME J 11(10)2391ndash2393
Gorlach K Shingaki R Morisaki H Hattori T 1994 Construction of eco-
collection of paddy field soil bacteria for population analysis J Gen
Appl Microbiol 40(6)509ndash517
Gourion B Rossignol M Vorholt JA Lindow SE 2006 A proteomic study
of Methylobacterium extorquens reveals a response regulator essential
for epiphytic growth wwwintegratedgenomicscom Accessed
November 26 2018
Green PN 2006 Methylobacterium In Dworkin M Falkow S Rosenberg
E Schleifer K Stackebrandt E editors The prokaryotes ndash a handbook
on the biology of bacteria New York Springer p 257ndash265
Hauruseu D Koblızek M 2012 Influence of light on carbon utilization in
aerobic anoxygenic phototrophs Appl Environ Microbiol
78(20)7414ndash7419
Katoh K Misawa K Kuma K Miyata T 2002 MAFFT a novel method for
rapid multiple sequence alignment based on fast Fourier transform
Nucleic Acids Res 30(14)3059ndash3066
Knief C et al 2012 Metaproteogenomic analysis of microbial communi-
ties in the phyllosphere and rhizosphere of rice ISME J
6(7)1378ndash1390
Knief C Frances L Vorholt JA 2010 Competitiveness of diverse methyl-
obacterium strains in the phyllosphere of arabidopsis thaliana and
identification of representative models including M extorquens
PA1 Microb Ecol 60(2)440ndash452
Kwak MJ et al 2014 Genome information of Methylobacterium oryzae a
plantndashprobiotic methylotroph in the phyllosphere PLoS One
9(9)e106704
Lindow SE Brandl MT 2003 Microbiology of the phyllosphere Appl
Environ Microbiol 69(4)1875ndash1883
Madsen AM Zervas A Tendal K Nielsen JL 2015 Microbial diversity in
bioaerosol samples causing ODTS compared to reference bioaerosol
samples as measured using Illumina sequencing and MALDI-TOF
Environ Res 140255ndash267
Martin M 2011 Cutadapt removes adapter sequences from high-
throughput sequencing reads EMBnet J 17(1)10ndash12
Marx CJ et al 2012 Complete genome sequences of six strains of the
genus Methylobacterium b J Bacteriol 194(17)4746
Noguchi T Hayashi H Shimada K Takaichi S Tasumi M 1992 In vivo
states and functions of carotenoids in an aerobic photosynthetic bac-
terium Erythrobacter longus Photosynth Res 31(1)21ndash30
Olm MR Brown CT Brooks B Banfield JF 2017 dRep a tool for fast and
accurate genomic comparisons that enables improved genome recov-
ery from metagenomes through de-replication ISME J
11(12)2864ndash2868
Overbeek R et al 2014 The SEED and the Rapid Annotation of microbial
genomes using Subsystems Technology (RAST) Nucl Acids Res
42(D1)D206ndashD214
Parks DH Imelfort M Skennerton CT Hugenholtz P Tyson GW 2015
CheckM assessing the quality of microbial genomes recovered from
isolates single cells and metagenomes Genome Res 25(7)1043ndash1055
Remus-Emsermann MNP et al 2014 Spatial distribution analyses of nat-
ural phyllosphere-colonizing bacteria on Arabidopsis thaliana revealed
by fluorescence in situ hybridization Environ Microbiol
16(7)2329ndash2340
Stamatakis A 2006 RAxML-VI-HPC maximum likelihood-based phyloge-
netic analyses with thousands of taxa and mixed models
Bioinformatics 22(21)2688ndash2690
UN 2017 World population prospects the 2017 revision Data Booklet
httpspopulationunorgwppPublicationsFilesWPP2019_DataBook-
letpdf
Vesth T Lagesen K Acar euroO Ussery D 2013 CMG-biotools a free work-
bench for basic comparative microbial genomics PLoS One
8(4)e60120
Vorholt JA 2012 Microbial life in the phyllosphere Nat Rev Microbiol
10(12)828ndash840
Wick RR Judd LM Gorrie CL Holt KE 2017 Unicycler resolving bacterial
genome assemblies from short and long sequencing reads PLoS
Comput Biol 13(6)e1005595
Wilson M Lindow SE 1994a Coexistence among epiphytic bacterial pop-
ulations mediated through nutritional resource partitioning Appl
Environ Microbiol 604468ndash77
Wilson M Lindow SE 1994b Ecological similarity and coexistence of epi-
phytic ice-nucleating (ice) pseudomonas syringae strains and a non-
ice-nucleating (ice) biological control agent Appl Environ Microbiol
60(9)3128ndash3137
Woodward FI Lomas MR 2004 Vegetation dynamics ndash simulating
responses to climatic change Biol Rev 79(3)643ndash670
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2907
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Xu J Gordon JI 2003 Honor thy symbionts Proc Natl Acad Sci USA
100(18)10452ndash10459
Yurkov V Csotonyi JT 2009 New light on aerobic anoxygenic photo-
trophs In Hunter NC Daldal F Thurnauer MC Beatty JT editors
The purple photosynthetic bacteria Dordrecht Springer thorn Business
Media BV p 31ndash55 doi101007978-1-4020-8815-5_3
Yurkov V Hughes E 2017 Aerobic Anoxygenic Phototrophs four
decades of mystery In Hallenbeck PC editor Modern topics in the
phototrophic prokaryotes environmental and applied aspects p
193ndash214
Yutin N et al 2009 BchY-based degenerate primers target all types of
anoxygenic photosynthetic bacteria in a single PCR Appl Environ
Microbiol 75(23)7556ndash7559
Zeng Y Feng F Medova H Dean J Koblizek M 2014 Functional
type 2 photosynthetic reaction centers found in the rare bacterial phy-
lum Gemmatimonadetes Proc Natl Acad Sci USA 111(21)7795ndash7800
Zheng Q et al 2011 Diverse arrangement of photosynthetic gene
clusters in aerobic anoxygenic phototrophic bacteria PLoS One
6(9)e25050
Associate editor Esperanza Martinez-Romero
Zervas et al GBE
2908 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
- evz204-TF1
- evz204-TF2
-
able to reproducibly distinguish different taxa and in cases
where reference MALDI-TOF spectra are available in
the database be able to assign Genus andor Species names
(Madsen et al 2015) We then proceeded with whole ge-
nome sequencing on the Illumina Nextseq platform and for
the unique genomes we also proceeded with Oxford
Nanopore Technologies sequencing to generate complete
genomes and plasmids
Following this multi-disciplinary approach that relies on
classical microbiological techniques (plating morphology
etc) biochemical analyses (MALDI-TOF MS) and two differ-
ent types of high-throughput whole genome sequencing we
were able to generate draft and complete genomes of novel
AAP bacteria from the wheat phyllosphere in a relatively quick
(weeks) and efficient manner without losing novel biological
information in the process of reducing redundancy
Bacteriochlorophyll-Containing AAP Bacteria in the WheatPhyllosphere
By investigating the presence of bacteriochlorophylls in the
wheat phyllosphere we were able to identify 21 unique strains
harboring them the majority of which belonged to the genus
Methylobacterium Members of this genus are ubiquitous in
nature and have been detected in both soil and plant surfaces
(Green 2006) Targeting the bchY (chlorophyllide reductase
subunit Y) and pufM (reaction centre subunit M) genes in
phyllosphere samples of different plants it was shown that
the Methylobacteriaceae (alpha-Proteobacteria) comprised
28 of the total pufM sequences from the amplicon sequenc-
ing analysis (Atamna-Ismaeel and Finkel 2012a) Our results
are thus in accordance to these previous studies As
Methylobacteriaceae are indigenous inhabitants of the phyllo-
sphere and have consistently been shown to harbor PGCs it
may be that these photosystems are essential to the
Methylobacteria that inhabit this niche Interestingly in their
study Atamna-Ismaeel et al had negative results from their
bchY amplicon sequencing approach suggesting the absence
of bacteria that contain RC1 (bacteriochlorophyll-a containing
reaction centers) type photosystems Our results show that all
isolates contain the complete cassette of chlorophyllide syn-
thase subunit encoding genes (bchB bchC bchF bchG bchI
bchL bchM bchN bchP bchX bchY bchZ) as well as both the
pufL and pufM genes (table 2) Testing the same primers (Yutin
et al 2009) in silico against our bchY sequences we had pos-
itive results for all of them (data not shown)
Patterns of loss of the carotenoid genes have been ob-
served in other alpha and gamma proteobacteria like in
Rhodobacterales and Sphingomonadales (Zheng et al
2011) The absence of crtA results in the inability to perform
the final step in the spheroidene pathway where hydroxy-
spheroidene is converted to hydroxyspheroidenone
Moreover the absence of crtC in Methylobacterium spp of
groups 3 and 4 completely affects both the spheroidene and
the spirilloxanthin pathways as its product is essential for the
first step of both metabolic pathways starting from neuro-
sporene and lycopene respectively This means that strains
FIG 4mdashWhole genome average nucleotide identity (ANI) cladogram
showing percentages of k-mer identities between the isolates Taxa with
the same color and above 90 identity threshold are considered to belong
in the same species Beijerinckia indica ATCC-9039 was used as outgroup
of the analyses (not defined a priori)
Zervas et al GBE
2902 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Tab
le2
Gen
eC
onte
nt
of
the
Photo
synth
esis
Gen
eC
lust
ers
Iden
tified
inth
e21
Bac
terial
Isola
tes
Bact
eri
och
loro
ph
yll
Syn
thase
Gen
es
Caro
ten
oid
Bio
syn
thesi
sG
en
es
Lig
ht-
Harv
est
ing
Co
mp
lex
Gen
es
Str
ain
acs
Fb
chB
bch
Cb
chF
bch
Gb
chI
bch
Lb
chM
bch
Nb
chP
bch
Xb
chY
bch
Zcr
tAcr
tBcr
tCcr
tDcr
tEcr
tFcr
tIcr
tKp
ucC
pu
fAp
ufB
pu
fCp
ufL
pu
fMp
uh
A
Rh
izo
biu
msp
W
L3
Ro
seo
mo
nas
sp
WL4
5
Als
ob
act
er
sp
WL4
Gro
up
1M
eth
ylo
bact
eri
um
sp
WL9
Meth
ylo
bact
eri
um
sp
WL1
9
Meth
ylo
bact
eri
um
sp
WL6
9
Gro
up
2M
eth
ylo
bact
eri
um
sp
WL1
Meth
ylo
bact
eri
um
sp
WL2
Meth
ylo
bact
eri
um
sp
WL7
Meth
ylo
bact
eri
um
sp
WL1
8
Meth
ylo
bact
eri
um
sp
WL6
4
Gro
up
3M
eth
ylo
bact
eri
um
sp
WL6
Meth
ylo
bact
eri
um
sp
WL3
0
Meth
ylo
bact
eri
um
sp
WL9
3
Meth
ylo
bact
eri
um
sp
WL1
16
Meth
ylo
bact
eri
um
sp
WL1
19
Gro
up
4M
eth
ylo
bact
eri
um
sp
WL8
Meth
ylo
bact
eri
um
sp
WL1
2
Meth
ylo
bact
eri
um
sp
WL1
03
Meth
ylo
bact
eri
um
sp
WL1
20
Meth
ylo
bact
eri
um
sp
WL1
22
NO
TEmdash
Gra
yb
oxe
sin
dic
ate
pre
sen
ceo
fg
en
es
an
dw
hit
eb
oxe
sin
dic
ate
ab
sen
ceo
fg
en
es
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2903
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
of these two groups rely on the zeaxanthin pathway to pro-
duce carotenoid pigments and most likely nostoxanthin and
erythroxanthin which however do not participate in the
light-harvesting process but rather have a photoprotection
role (Noguchi et al 1992 Yurkov and Csotonyi 2009)
Phylogenetic Observations
In our study we sequenced and analyzed 21 aerobic anoxy-
genic photosynthetic bacteria most of which belonged to the
genus Methylobacterium the presence of which in the
environment has already been discussed previously
Methylobacterium spp are facultative methylotrophs and
can utilize a variety of C1 C2 C3 and C4 compounds
Such compounds are often found in the phyllosphere due
to plant metabolism (Vorholt 2012 Remus-Emsermann
et al 2014) Methylobacterium spp show a wide range in
chromosome size and number of plasmids they contain
(Marx et al 2012) The genus is represented well in
GenBank with 100 complete or draft genomes (last accessed
April 2019) However most of the sequences do not have
FIG 5mdashArchitecture of the different photosynthetic gene clusters Arrows indicate the orientation of the genes Sizes are not representatives of gene
length Only the genes shared by all PGCs were shown Genes that may be missing from some strains are denoted with parentheses on the figurersquos legend
Table 3
General Descriptive Statistics of the Four Complete AAP Bacterial Genomes
Taxon Genome Size (bp) GC Number of Genes Plasmids Plasmids Size (bp) GC
Rhizobium sp WL3 4568855 6160 4450 1 700631 586
mdash mdash mdash 2 85350 575
Roseomonas sp WL45 4533887 708 5042 1 262815 655
1011854 707 1138 2 240556 662
mdash mdash mdash 3 152922 658
mdash mdash mdash 4 85279 646
mdash mdash mdash 5 84774 661
mdash mdash mdash 6 83273 656
mdash mdash mdash 7 65860 648
Alsobacter sp WL4 5405668 679 5013 1 27178 620
mdash mdash mdash 2 9451 606
Methylobacterium sp WL1 6215463 691 6431 1 35731 638
Zervas et al GBE
2904 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
species affiliation (fig 4) Thus we were not able to assign any
of the 18 isolates to specific Methylobacterium species apart
perhaps from group 2 (WL1 WL2 WL7 WL18 WL64) which
is placed as sister to M brachiatum in the MASH-ANI analysis
Even in this case though M pseudosasicola BL36 is also
placed in the same clade as M brachiatum which raises the
question of proper nomenclature This is further evidenced in
the same tree (cyanide clade below) where seven species
share gt90 of their genome For the remaining isolates
WL19 is closely affiliated to other Methylobacterium strains
that have not been identified A similar picture is shown for
WL69 On the other hand WL19 appears to be a novel spe-
cies with unique genomic features and so do the strains from
groups 3 and 4 (red clade) Thus despite how ubiquitous
Methylobacterium is in nature it is clear that part of its diver-
gence is still missing Moreover a critical evaluation of the
current nomenclature of the genus would be beneficial based
on the whole genome sequences because closely related
genomes (MASH-ANI identitygt 97) have been given differ-
ent species affiliations
Apart from this genus we also identified a Rhizobium sp
a Roseomonas sp and a novel Alsobacter strain (WL4) that
belongs in the Methylocystaceae a family of methanotroph
bacteria This is the first time that AAP bacteria from this
family have been isolated and their complete genomes assem-
bled Its closely related Alsobacter spp have been shown to
exhibit high Thallium tolerance (Bao et al 2006) and it will be
interesting to identify heavy-metal resistance genes in non-
heavy-metal polluted phyllosphere samples It is worth men-
tioning that if WL4 contains heavy-metal resistance genes on
top of the PGC this will have been shown for the first time
and its potential to bioremediate heavy-metal contaminated
areas if applied as plant growth promotion and environmental
remediation agent is of high value and needs to further be
explored
Molecular Evolution of the PGCs
For all gene trees of the PGC we observed that Rhizobium sp
WL3 evolves faster than the other isolates followed by
Table 4
Substitutions Per Site in the Different Phylogenetic Analyses
Strain 16S acsF bchL bchY crtB pufL pufMsuper-matrix
CheckM
Roseomonas sp WL45 0096 0147 0100 0235 0250 0095 0132 0160 0184
Rhizobium sp WL3 0184 0622 0767 0690 1370 0343 0525 0580 0372
Alsobacter sp WL4 0151 0462 0233 0490 0870 0258 0334 0380 0403
Methylobacterium sp WL9 0226 0418 0372 0420 0670 0270 0319 0364 0477
Methylobacterium sp WL19 0221 0413 0300 0460 0600 0233 0346 0364 0468
Methylobacterium sp WL69 0231 0373 0333 0470 0620 0260 0358 0368 0468
Methylobacterium sp WL1 0233 0422 0344 0485 0480 0280 0385 0380 0459
Methylobacterium sp WL2 0233 0422 0344 0485 0480 0280 0385 0380 0459
Methylobacterium sp WL7 0238 0427 0344 0480 0480 0278 0385 0380 0459
Methylobacterium sp WL18 0238 0427 0344 0480 0480 0280 0385 0380 0459
Methylobacterium sp WL64 0233 0422 0344 0480 0485 0280 0385 0380 0459
Methylobacterium sp WL6 0199 0378 0256 0425 0680 0208 0311 0340 0455
Methylobacterium sp WL30 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL93 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL116 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL119 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL8 0188 0373 0258 0430 0680 0230 0358 0348 0459
Methylobacterium sp WL12 0188 0373 0256 0430 0680 0230 0358 0348 0459
Methylobacterium sp WL103 0188 0373 0256 0430 0850 0230 0358 0364 0459
Methylobacterium sp WL120 0188 0373 0258 0430 0680 0233 0358 0348 0459
Methylobacterium sp WL122 0188 0373 0256 0430 0680 0230 0358 0348 0459
Susbstitutions per site
Gro
up1
Gro
up 2
Gro
up 3
Gro
up 4
NOTEmdashAll values correspond to distances from the root where Roseomonas sp WL45 was chosen as outgroup for all the alignments
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2905
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Alsobacter sp WL4 Both these isolates also show much
higher substitution rates compared with their respective 16S
genes and core proteomes (checkM) which indicates that the
PGCs are under relaxed selection pressure In all instances the
phylogenetic trees constructed showed the same topology
and the same grouping of the Methylobacterium isolates
which suggests that the PGCs were incorporated in the
genomes of these strains relatively early in their evolution If
the PGC was free to move laterally then we would have ob-
served very different phylogenetic trees for its genes com-
pared with the 16S and core proteome analyses
Interestingly these observations do not reflect on the archi-
tecture of the PGC where Roseomonas sp WL45 and
Rhizobium sp WL3 contain the PGC in one continuous re-
gion whereas Alsobacter sp WL4 and Methylobacterium sp
WL1 have it in two pieces
For the 18 Methylobacterium isolates group 1 and 2
strains (WL9 WL19 WL69 and WL1 WL2 WL7 WL18
WL64) evolve consistently at different rates compared
with the groups 3 and 4 It appears that there is relatively
strong selection pressure on bchL pufL and pufM which
in all cases are more similar to each other and also show
slower molecular evolution rates compared with ascF
bchY and crtB which in all Methylobacterium isolates
evolve at least twice as fast compared with the 16S
gene (table 4) They also show fewer substitutions per
site compared with the core proteome analysis This result
partially contrasted with previously adopted methods of
using bchY as a marker gene for RC1 clusters (eg
Atamna-Ismaeel and Finkel 2012a) On the other hand
the use of pufM appears more sensible It would how-
ever be more suitable to investigate the possibility of us-
ing other genes for metagenome amplicon sequencing
approaches such as pufL or bchL which show a smaller
degree of divergence in different genera thus universal
primers might be more successful in detecting a wide va-
riety of AAP bacteria in environmental samples
Interestingly while Groups 1 and 2 isolates exhibit higher
substitution rates of the PGC genes compared with
Groups 3 and 4 this is not the case for crtB which evolves
a lot slower and also pufM which shows similar molec-
ular evolution in the genus Overall it seems that the
PGCs were incorporated in ancestral strains that later di-
verged to give rise to the different AAP taxa This is fur-
ther corroborated by the fact that almost all phylogenetic
trees (eg fig 5) are consistent and in agreement with
both the 16S phylogenetic tree (fig 1) as well as the phy-
logenomic tree (fig 2) If the genes of the PGCs had been
transferred to these AAP several times there would have
been significant differences both in the sequence level
and the phylogenetic analysis of the different genes
which would be expected to result in different topologies
and diverse distances from the root even for closelymdash
based on 16S analysismdashrelated strains
The Significance of AAP Bacteria in Wheat Phyllosphere
In this study we isolated for the first time a variety of AAP
bacteria that are present in the wheat phyllosphere adding to
the metagenomics knowledge provided earlier from five other
plants (Atamna-Ismaeel and Finkel 2012a 2012b) The phyllo-
sphere is a rather harsh environment with little water avail-
ability scarce resources very few suitable locations extreme
temperature difference between day and night and high
dosages of UV radiation (Vorholt 2012) AAP bacteria have
the ability to utilize light in photophosphorylation processes
through which they produce ATP and at the same time funnel
the few available nutrients in other metabolic pathways
(Yurkov and Hughes 2017) This has been shown to give
them an edge in vitro after exposure to light compared with
nonAAP bacteria (Ferrera et al 2011) Considering that phyl-
losphere bacteria protect the plant from pathogenic microbes
by secreting metabolites and by occupying the available
niches being able to persevere and outgrow other bacteria
gives a significant edge to AAP taxa and probably points
toward the suggestion of positive selection of such bacteria
in their phyllosphere by the plants themselves as they appear
to be absent from the soil (Atamna-Ismaeel and Finkel
2012b) At this point it is not clear exactly how AAP bacteria
benefit the plant especially since the phyllosphere is a com-
plex habitat with an ever-changing harsh environment In the
past years great efforts have been carried out on how to
improve crop production by studying the rhizosphere It has
also been suggested that the phyllosphere communities play a
significant role in plant health and growth which ultimately
leads to increased yields (Lindow and Brandl 2003) Given the
current projections of world population growth and the sup-
ply of food (UN 2017) it becomes evident that more in-depth
studies investigating the phyllosphere of plants and especially
crops and identifying the underlying drivers of the bacterial
communities and their interactions with their plant hosts is of
paramount importance Thus by isolating maintaining and
analyzing the genomes of phyllosphere isolates we may iden-
tify strains that may prove useful as plant growth promoting
agents in future field trials investigating the effect of phyllo-
sphere inoculation on plant growth health and yield
Supplementary Material
Supplementary data are available at Genome Biology and
Evolution online
Acknowledgments
The authors would like to thank Tue Sparholt Jorgensen for
collecting the wheat sample The authors also want to thank
Tina Thane (AU) Tanja Begovic (AU) and Margit Frederiksen
(NRCWE) for capable laboratory assistance We thank Michal
Koblızek for providing the initial model of the colony infrared
imaging system This work was funded by a Villum
Zervas et al GBE
2906 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Experiment grant (no 17601) and a Marie Skłodowska-Curie
AIAS-COFUND fellowship (EU-FP7 program Grant
Agreement No 609033) awarded to YZ
Author Contributions
AZ YZ LHH designed the study YZ isolated the strains
performed colony infrared imaging and extracted DNA for
Illumina sequencing YZ and AMM performed the MALDI-
TOF MS analysis AZ extracted DNA and performed ONT
Sequencing AZ carried out the bioinformatics and phyloge-
netic analyses AZ and YZ drafted the manuscript AZ
YZ AMM and LHH revised the manuscript for
intellectual content All authors read and approved the
manuscript
Literature Cited
Atamna-Ismaeel N Finkel O 2012a Bacterial anoxygenic photosynthesis
on plant leaf surfaces Environ Microbiol Rep 4(2)209ndash216
Atamna-Ismaeel N Finkel OM 2012b Microbial rhodopsins on leaf sur-
faces of terrestrial plants Environ Microbiol 14(1)140ndash146
Aziz RK et al 2008 The RAST Server rapid annotations using subsystems
technology BMC Genomics 9(1)75
Bankevich A et al 2012 SPAdes a new genome assembly algorithm and
its applications to single-cell sequencing J Comput Biol
19(5)455ndash477
Bao Z Sato Y Kubota M Ohta H 2006 Isolation and characterization of
thallium-tolerant bacteria from heavy metal-polluted river sediment
and non-polluted soils Microb Environ 21(4)251ndash260
Beatie GA 2002 Leaf surface waxes and the process of leaf colonization
by microorganisms In Lindow SE Hecht-Poinar EI Elliott VJ editors
Phyllosphere Microbiology St Paul (USA) APS Press pp 3ndash26
Beattie GA Lindow SE 1999 Bacterial colonization of leaves a spectrum
of strategies Phytopathology 89(5)353ndash359
Brettin T et al 2015 RASTtk a modular and extensible implementation of
the RAST algorithm for building custom annotation pipelines and an-
notating batches of genomes Sci Rep 58365
Carvalho SD Castillo JA 2018 Influence of light on plantndashphyllosphere
interaction Front Plant Sci 91482
Ferrera I Gasol JM Sebastian M Hojerova E Koblızek M 2011
Comparison of growth rates of aerobic anoxygenic phototrophic bac-
teria and other bacterioplankton groups in coastal Mediterranean wa-
ters Appl Environ Microbiol 77(21)7451ndash7458
Ferrera I Sanchez O Kolarova E Koblızek M Gasol JM 2017 Light
enhances the growth rates of natural populations of aerobic anoxy-
genic phototrophic bacteria ISME J 11(10)2391ndash2393
Gorlach K Shingaki R Morisaki H Hattori T 1994 Construction of eco-
collection of paddy field soil bacteria for population analysis J Gen
Appl Microbiol 40(6)509ndash517
Gourion B Rossignol M Vorholt JA Lindow SE 2006 A proteomic study
of Methylobacterium extorquens reveals a response regulator essential
for epiphytic growth wwwintegratedgenomicscom Accessed
November 26 2018
Green PN 2006 Methylobacterium In Dworkin M Falkow S Rosenberg
E Schleifer K Stackebrandt E editors The prokaryotes ndash a handbook
on the biology of bacteria New York Springer p 257ndash265
Hauruseu D Koblızek M 2012 Influence of light on carbon utilization in
aerobic anoxygenic phototrophs Appl Environ Microbiol
78(20)7414ndash7419
Katoh K Misawa K Kuma K Miyata T 2002 MAFFT a novel method for
rapid multiple sequence alignment based on fast Fourier transform
Nucleic Acids Res 30(14)3059ndash3066
Knief C et al 2012 Metaproteogenomic analysis of microbial communi-
ties in the phyllosphere and rhizosphere of rice ISME J
6(7)1378ndash1390
Knief C Frances L Vorholt JA 2010 Competitiveness of diverse methyl-
obacterium strains in the phyllosphere of arabidopsis thaliana and
identification of representative models including M extorquens
PA1 Microb Ecol 60(2)440ndash452
Kwak MJ et al 2014 Genome information of Methylobacterium oryzae a
plantndashprobiotic methylotroph in the phyllosphere PLoS One
9(9)e106704
Lindow SE Brandl MT 2003 Microbiology of the phyllosphere Appl
Environ Microbiol 69(4)1875ndash1883
Madsen AM Zervas A Tendal K Nielsen JL 2015 Microbial diversity in
bioaerosol samples causing ODTS compared to reference bioaerosol
samples as measured using Illumina sequencing and MALDI-TOF
Environ Res 140255ndash267
Martin M 2011 Cutadapt removes adapter sequences from high-
throughput sequencing reads EMBnet J 17(1)10ndash12
Marx CJ et al 2012 Complete genome sequences of six strains of the
genus Methylobacterium b J Bacteriol 194(17)4746
Noguchi T Hayashi H Shimada K Takaichi S Tasumi M 1992 In vivo
states and functions of carotenoids in an aerobic photosynthetic bac-
terium Erythrobacter longus Photosynth Res 31(1)21ndash30
Olm MR Brown CT Brooks B Banfield JF 2017 dRep a tool for fast and
accurate genomic comparisons that enables improved genome recov-
ery from metagenomes through de-replication ISME J
11(12)2864ndash2868
Overbeek R et al 2014 The SEED and the Rapid Annotation of microbial
genomes using Subsystems Technology (RAST) Nucl Acids Res
42(D1)D206ndashD214
Parks DH Imelfort M Skennerton CT Hugenholtz P Tyson GW 2015
CheckM assessing the quality of microbial genomes recovered from
isolates single cells and metagenomes Genome Res 25(7)1043ndash1055
Remus-Emsermann MNP et al 2014 Spatial distribution analyses of nat-
ural phyllosphere-colonizing bacteria on Arabidopsis thaliana revealed
by fluorescence in situ hybridization Environ Microbiol
16(7)2329ndash2340
Stamatakis A 2006 RAxML-VI-HPC maximum likelihood-based phyloge-
netic analyses with thousands of taxa and mixed models
Bioinformatics 22(21)2688ndash2690
UN 2017 World population prospects the 2017 revision Data Booklet
httpspopulationunorgwppPublicationsFilesWPP2019_DataBook-
letpdf
Vesth T Lagesen K Acar euroO Ussery D 2013 CMG-biotools a free work-
bench for basic comparative microbial genomics PLoS One
8(4)e60120
Vorholt JA 2012 Microbial life in the phyllosphere Nat Rev Microbiol
10(12)828ndash840
Wick RR Judd LM Gorrie CL Holt KE 2017 Unicycler resolving bacterial
genome assemblies from short and long sequencing reads PLoS
Comput Biol 13(6)e1005595
Wilson M Lindow SE 1994a Coexistence among epiphytic bacterial pop-
ulations mediated through nutritional resource partitioning Appl
Environ Microbiol 604468ndash77
Wilson M Lindow SE 1994b Ecological similarity and coexistence of epi-
phytic ice-nucleating (ice) pseudomonas syringae strains and a non-
ice-nucleating (ice) biological control agent Appl Environ Microbiol
60(9)3128ndash3137
Woodward FI Lomas MR 2004 Vegetation dynamics ndash simulating
responses to climatic change Biol Rev 79(3)643ndash670
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2907
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Xu J Gordon JI 2003 Honor thy symbionts Proc Natl Acad Sci USA
100(18)10452ndash10459
Yurkov V Csotonyi JT 2009 New light on aerobic anoxygenic photo-
trophs In Hunter NC Daldal F Thurnauer MC Beatty JT editors
The purple photosynthetic bacteria Dordrecht Springer thorn Business
Media BV p 31ndash55 doi101007978-1-4020-8815-5_3
Yurkov V Hughes E 2017 Aerobic Anoxygenic Phototrophs four
decades of mystery In Hallenbeck PC editor Modern topics in the
phototrophic prokaryotes environmental and applied aspects p
193ndash214
Yutin N et al 2009 BchY-based degenerate primers target all types of
anoxygenic photosynthetic bacteria in a single PCR Appl Environ
Microbiol 75(23)7556ndash7559
Zeng Y Feng F Medova H Dean J Koblizek M 2014 Functional
type 2 photosynthetic reaction centers found in the rare bacterial phy-
lum Gemmatimonadetes Proc Natl Acad Sci USA 111(21)7795ndash7800
Zheng Q et al 2011 Diverse arrangement of photosynthetic gene
clusters in aerobic anoxygenic phototrophic bacteria PLoS One
6(9)e25050
Associate editor Esperanza Martinez-Romero
Zervas et al GBE
2908 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
- evz204-TF1
- evz204-TF2
-
Tab
le2
Gen
eC
onte
nt
of
the
Photo
synth
esis
Gen
eC
lust
ers
Iden
tified
inth
e21
Bac
terial
Isola
tes
Bact
eri
och
loro
ph
yll
Syn
thase
Gen
es
Caro
ten
oid
Bio
syn
thesi
sG
en
es
Lig
ht-
Harv
est
ing
Co
mp
lex
Gen
es
Str
ain
acs
Fb
chB
bch
Cb
chF
bch
Gb
chI
bch
Lb
chM
bch
Nb
chP
bch
Xb
chY
bch
Zcr
tAcr
tBcr
tCcr
tDcr
tEcr
tFcr
tIcr
tKp
ucC
pu
fAp
ufB
pu
fCp
ufL
pu
fMp
uh
A
Rh
izo
biu
msp
W
L3
Ro
seo
mo
nas
sp
WL4
5
Als
ob
act
er
sp
WL4
Gro
up
1M
eth
ylo
bact
eri
um
sp
WL9
Meth
ylo
bact
eri
um
sp
WL1
9
Meth
ylo
bact
eri
um
sp
WL6
9
Gro
up
2M
eth
ylo
bact
eri
um
sp
WL1
Meth
ylo
bact
eri
um
sp
WL2
Meth
ylo
bact
eri
um
sp
WL7
Meth
ylo
bact
eri
um
sp
WL1
8
Meth
ylo
bact
eri
um
sp
WL6
4
Gro
up
3M
eth
ylo
bact
eri
um
sp
WL6
Meth
ylo
bact
eri
um
sp
WL3
0
Meth
ylo
bact
eri
um
sp
WL9
3
Meth
ylo
bact
eri
um
sp
WL1
16
Meth
ylo
bact
eri
um
sp
WL1
19
Gro
up
4M
eth
ylo
bact
eri
um
sp
WL8
Meth
ylo
bact
eri
um
sp
WL1
2
Meth
ylo
bact
eri
um
sp
WL1
03
Meth
ylo
bact
eri
um
sp
WL1
20
Meth
ylo
bact
eri
um
sp
WL1
22
NO
TEmdash
Gra
yb
oxe
sin
dic
ate
pre
sen
ceo
fg
en
es
an
dw
hit
eb
oxe
sin
dic
ate
ab
sen
ceo
fg
en
es
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2903
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
of these two groups rely on the zeaxanthin pathway to pro-
duce carotenoid pigments and most likely nostoxanthin and
erythroxanthin which however do not participate in the
light-harvesting process but rather have a photoprotection
role (Noguchi et al 1992 Yurkov and Csotonyi 2009)
Phylogenetic Observations
In our study we sequenced and analyzed 21 aerobic anoxy-
genic photosynthetic bacteria most of which belonged to the
genus Methylobacterium the presence of which in the
environment has already been discussed previously
Methylobacterium spp are facultative methylotrophs and
can utilize a variety of C1 C2 C3 and C4 compounds
Such compounds are often found in the phyllosphere due
to plant metabolism (Vorholt 2012 Remus-Emsermann
et al 2014) Methylobacterium spp show a wide range in
chromosome size and number of plasmids they contain
(Marx et al 2012) The genus is represented well in
GenBank with 100 complete or draft genomes (last accessed
April 2019) However most of the sequences do not have
FIG 5mdashArchitecture of the different photosynthetic gene clusters Arrows indicate the orientation of the genes Sizes are not representatives of gene
length Only the genes shared by all PGCs were shown Genes that may be missing from some strains are denoted with parentheses on the figurersquos legend
Table 3
General Descriptive Statistics of the Four Complete AAP Bacterial Genomes
Taxon Genome Size (bp) GC Number of Genes Plasmids Plasmids Size (bp) GC
Rhizobium sp WL3 4568855 6160 4450 1 700631 586
mdash mdash mdash 2 85350 575
Roseomonas sp WL45 4533887 708 5042 1 262815 655
1011854 707 1138 2 240556 662
mdash mdash mdash 3 152922 658
mdash mdash mdash 4 85279 646
mdash mdash mdash 5 84774 661
mdash mdash mdash 6 83273 656
mdash mdash mdash 7 65860 648
Alsobacter sp WL4 5405668 679 5013 1 27178 620
mdash mdash mdash 2 9451 606
Methylobacterium sp WL1 6215463 691 6431 1 35731 638
Zervas et al GBE
2904 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
species affiliation (fig 4) Thus we were not able to assign any
of the 18 isolates to specific Methylobacterium species apart
perhaps from group 2 (WL1 WL2 WL7 WL18 WL64) which
is placed as sister to M brachiatum in the MASH-ANI analysis
Even in this case though M pseudosasicola BL36 is also
placed in the same clade as M brachiatum which raises the
question of proper nomenclature This is further evidenced in
the same tree (cyanide clade below) where seven species
share gt90 of their genome For the remaining isolates
WL19 is closely affiliated to other Methylobacterium strains
that have not been identified A similar picture is shown for
WL69 On the other hand WL19 appears to be a novel spe-
cies with unique genomic features and so do the strains from
groups 3 and 4 (red clade) Thus despite how ubiquitous
Methylobacterium is in nature it is clear that part of its diver-
gence is still missing Moreover a critical evaluation of the
current nomenclature of the genus would be beneficial based
on the whole genome sequences because closely related
genomes (MASH-ANI identitygt 97) have been given differ-
ent species affiliations
Apart from this genus we also identified a Rhizobium sp
a Roseomonas sp and a novel Alsobacter strain (WL4) that
belongs in the Methylocystaceae a family of methanotroph
bacteria This is the first time that AAP bacteria from this
family have been isolated and their complete genomes assem-
bled Its closely related Alsobacter spp have been shown to
exhibit high Thallium tolerance (Bao et al 2006) and it will be
interesting to identify heavy-metal resistance genes in non-
heavy-metal polluted phyllosphere samples It is worth men-
tioning that if WL4 contains heavy-metal resistance genes on
top of the PGC this will have been shown for the first time
and its potential to bioremediate heavy-metal contaminated
areas if applied as plant growth promotion and environmental
remediation agent is of high value and needs to further be
explored
Molecular Evolution of the PGCs
For all gene trees of the PGC we observed that Rhizobium sp
WL3 evolves faster than the other isolates followed by
Table 4
Substitutions Per Site in the Different Phylogenetic Analyses
Strain 16S acsF bchL bchY crtB pufL pufMsuper-matrix
CheckM
Roseomonas sp WL45 0096 0147 0100 0235 0250 0095 0132 0160 0184
Rhizobium sp WL3 0184 0622 0767 0690 1370 0343 0525 0580 0372
Alsobacter sp WL4 0151 0462 0233 0490 0870 0258 0334 0380 0403
Methylobacterium sp WL9 0226 0418 0372 0420 0670 0270 0319 0364 0477
Methylobacterium sp WL19 0221 0413 0300 0460 0600 0233 0346 0364 0468
Methylobacterium sp WL69 0231 0373 0333 0470 0620 0260 0358 0368 0468
Methylobacterium sp WL1 0233 0422 0344 0485 0480 0280 0385 0380 0459
Methylobacterium sp WL2 0233 0422 0344 0485 0480 0280 0385 0380 0459
Methylobacterium sp WL7 0238 0427 0344 0480 0480 0278 0385 0380 0459
Methylobacterium sp WL18 0238 0427 0344 0480 0480 0280 0385 0380 0459
Methylobacterium sp WL64 0233 0422 0344 0480 0485 0280 0385 0380 0459
Methylobacterium sp WL6 0199 0378 0256 0425 0680 0208 0311 0340 0455
Methylobacterium sp WL30 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL93 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL116 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL119 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL8 0188 0373 0258 0430 0680 0230 0358 0348 0459
Methylobacterium sp WL12 0188 0373 0256 0430 0680 0230 0358 0348 0459
Methylobacterium sp WL103 0188 0373 0256 0430 0850 0230 0358 0364 0459
Methylobacterium sp WL120 0188 0373 0258 0430 0680 0233 0358 0348 0459
Methylobacterium sp WL122 0188 0373 0256 0430 0680 0230 0358 0348 0459
Susbstitutions per site
Gro
up1
Gro
up 2
Gro
up 3
Gro
up 4
NOTEmdashAll values correspond to distances from the root where Roseomonas sp WL45 was chosen as outgroup for all the alignments
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2905
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Alsobacter sp WL4 Both these isolates also show much
higher substitution rates compared with their respective 16S
genes and core proteomes (checkM) which indicates that the
PGCs are under relaxed selection pressure In all instances the
phylogenetic trees constructed showed the same topology
and the same grouping of the Methylobacterium isolates
which suggests that the PGCs were incorporated in the
genomes of these strains relatively early in their evolution If
the PGC was free to move laterally then we would have ob-
served very different phylogenetic trees for its genes com-
pared with the 16S and core proteome analyses
Interestingly these observations do not reflect on the archi-
tecture of the PGC where Roseomonas sp WL45 and
Rhizobium sp WL3 contain the PGC in one continuous re-
gion whereas Alsobacter sp WL4 and Methylobacterium sp
WL1 have it in two pieces
For the 18 Methylobacterium isolates group 1 and 2
strains (WL9 WL19 WL69 and WL1 WL2 WL7 WL18
WL64) evolve consistently at different rates compared
with the groups 3 and 4 It appears that there is relatively
strong selection pressure on bchL pufL and pufM which
in all cases are more similar to each other and also show
slower molecular evolution rates compared with ascF
bchY and crtB which in all Methylobacterium isolates
evolve at least twice as fast compared with the 16S
gene (table 4) They also show fewer substitutions per
site compared with the core proteome analysis This result
partially contrasted with previously adopted methods of
using bchY as a marker gene for RC1 clusters (eg
Atamna-Ismaeel and Finkel 2012a) On the other hand
the use of pufM appears more sensible It would how-
ever be more suitable to investigate the possibility of us-
ing other genes for metagenome amplicon sequencing
approaches such as pufL or bchL which show a smaller
degree of divergence in different genera thus universal
primers might be more successful in detecting a wide va-
riety of AAP bacteria in environmental samples
Interestingly while Groups 1 and 2 isolates exhibit higher
substitution rates of the PGC genes compared with
Groups 3 and 4 this is not the case for crtB which evolves
a lot slower and also pufM which shows similar molec-
ular evolution in the genus Overall it seems that the
PGCs were incorporated in ancestral strains that later di-
verged to give rise to the different AAP taxa This is fur-
ther corroborated by the fact that almost all phylogenetic
trees (eg fig 5) are consistent and in agreement with
both the 16S phylogenetic tree (fig 1) as well as the phy-
logenomic tree (fig 2) If the genes of the PGCs had been
transferred to these AAP several times there would have
been significant differences both in the sequence level
and the phylogenetic analysis of the different genes
which would be expected to result in different topologies
and diverse distances from the root even for closelymdash
based on 16S analysismdashrelated strains
The Significance of AAP Bacteria in Wheat Phyllosphere
In this study we isolated for the first time a variety of AAP
bacteria that are present in the wheat phyllosphere adding to
the metagenomics knowledge provided earlier from five other
plants (Atamna-Ismaeel and Finkel 2012a 2012b) The phyllo-
sphere is a rather harsh environment with little water avail-
ability scarce resources very few suitable locations extreme
temperature difference between day and night and high
dosages of UV radiation (Vorholt 2012) AAP bacteria have
the ability to utilize light in photophosphorylation processes
through which they produce ATP and at the same time funnel
the few available nutrients in other metabolic pathways
(Yurkov and Hughes 2017) This has been shown to give
them an edge in vitro after exposure to light compared with
nonAAP bacteria (Ferrera et al 2011) Considering that phyl-
losphere bacteria protect the plant from pathogenic microbes
by secreting metabolites and by occupying the available
niches being able to persevere and outgrow other bacteria
gives a significant edge to AAP taxa and probably points
toward the suggestion of positive selection of such bacteria
in their phyllosphere by the plants themselves as they appear
to be absent from the soil (Atamna-Ismaeel and Finkel
2012b) At this point it is not clear exactly how AAP bacteria
benefit the plant especially since the phyllosphere is a com-
plex habitat with an ever-changing harsh environment In the
past years great efforts have been carried out on how to
improve crop production by studying the rhizosphere It has
also been suggested that the phyllosphere communities play a
significant role in plant health and growth which ultimately
leads to increased yields (Lindow and Brandl 2003) Given the
current projections of world population growth and the sup-
ply of food (UN 2017) it becomes evident that more in-depth
studies investigating the phyllosphere of plants and especially
crops and identifying the underlying drivers of the bacterial
communities and their interactions with their plant hosts is of
paramount importance Thus by isolating maintaining and
analyzing the genomes of phyllosphere isolates we may iden-
tify strains that may prove useful as plant growth promoting
agents in future field trials investigating the effect of phyllo-
sphere inoculation on plant growth health and yield
Supplementary Material
Supplementary data are available at Genome Biology and
Evolution online
Acknowledgments
The authors would like to thank Tue Sparholt Jorgensen for
collecting the wheat sample The authors also want to thank
Tina Thane (AU) Tanja Begovic (AU) and Margit Frederiksen
(NRCWE) for capable laboratory assistance We thank Michal
Koblızek for providing the initial model of the colony infrared
imaging system This work was funded by a Villum
Zervas et al GBE
2906 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Experiment grant (no 17601) and a Marie Skłodowska-Curie
AIAS-COFUND fellowship (EU-FP7 program Grant
Agreement No 609033) awarded to YZ
Author Contributions
AZ YZ LHH designed the study YZ isolated the strains
performed colony infrared imaging and extracted DNA for
Illumina sequencing YZ and AMM performed the MALDI-
TOF MS analysis AZ extracted DNA and performed ONT
Sequencing AZ carried out the bioinformatics and phyloge-
netic analyses AZ and YZ drafted the manuscript AZ
YZ AMM and LHH revised the manuscript for
intellectual content All authors read and approved the
manuscript
Literature Cited
Atamna-Ismaeel N Finkel O 2012a Bacterial anoxygenic photosynthesis
on plant leaf surfaces Environ Microbiol Rep 4(2)209ndash216
Atamna-Ismaeel N Finkel OM 2012b Microbial rhodopsins on leaf sur-
faces of terrestrial plants Environ Microbiol 14(1)140ndash146
Aziz RK et al 2008 The RAST Server rapid annotations using subsystems
technology BMC Genomics 9(1)75
Bankevich A et al 2012 SPAdes a new genome assembly algorithm and
its applications to single-cell sequencing J Comput Biol
19(5)455ndash477
Bao Z Sato Y Kubota M Ohta H 2006 Isolation and characterization of
thallium-tolerant bacteria from heavy metal-polluted river sediment
and non-polluted soils Microb Environ 21(4)251ndash260
Beatie GA 2002 Leaf surface waxes and the process of leaf colonization
by microorganisms In Lindow SE Hecht-Poinar EI Elliott VJ editors
Phyllosphere Microbiology St Paul (USA) APS Press pp 3ndash26
Beattie GA Lindow SE 1999 Bacterial colonization of leaves a spectrum
of strategies Phytopathology 89(5)353ndash359
Brettin T et al 2015 RASTtk a modular and extensible implementation of
the RAST algorithm for building custom annotation pipelines and an-
notating batches of genomes Sci Rep 58365
Carvalho SD Castillo JA 2018 Influence of light on plantndashphyllosphere
interaction Front Plant Sci 91482
Ferrera I Gasol JM Sebastian M Hojerova E Koblızek M 2011
Comparison of growth rates of aerobic anoxygenic phototrophic bac-
teria and other bacterioplankton groups in coastal Mediterranean wa-
ters Appl Environ Microbiol 77(21)7451ndash7458
Ferrera I Sanchez O Kolarova E Koblızek M Gasol JM 2017 Light
enhances the growth rates of natural populations of aerobic anoxy-
genic phototrophic bacteria ISME J 11(10)2391ndash2393
Gorlach K Shingaki R Morisaki H Hattori T 1994 Construction of eco-
collection of paddy field soil bacteria for population analysis J Gen
Appl Microbiol 40(6)509ndash517
Gourion B Rossignol M Vorholt JA Lindow SE 2006 A proteomic study
of Methylobacterium extorquens reveals a response regulator essential
for epiphytic growth wwwintegratedgenomicscom Accessed
November 26 2018
Green PN 2006 Methylobacterium In Dworkin M Falkow S Rosenberg
E Schleifer K Stackebrandt E editors The prokaryotes ndash a handbook
on the biology of bacteria New York Springer p 257ndash265
Hauruseu D Koblızek M 2012 Influence of light on carbon utilization in
aerobic anoxygenic phototrophs Appl Environ Microbiol
78(20)7414ndash7419
Katoh K Misawa K Kuma K Miyata T 2002 MAFFT a novel method for
rapid multiple sequence alignment based on fast Fourier transform
Nucleic Acids Res 30(14)3059ndash3066
Knief C et al 2012 Metaproteogenomic analysis of microbial communi-
ties in the phyllosphere and rhizosphere of rice ISME J
6(7)1378ndash1390
Knief C Frances L Vorholt JA 2010 Competitiveness of diverse methyl-
obacterium strains in the phyllosphere of arabidopsis thaliana and
identification of representative models including M extorquens
PA1 Microb Ecol 60(2)440ndash452
Kwak MJ et al 2014 Genome information of Methylobacterium oryzae a
plantndashprobiotic methylotroph in the phyllosphere PLoS One
9(9)e106704
Lindow SE Brandl MT 2003 Microbiology of the phyllosphere Appl
Environ Microbiol 69(4)1875ndash1883
Madsen AM Zervas A Tendal K Nielsen JL 2015 Microbial diversity in
bioaerosol samples causing ODTS compared to reference bioaerosol
samples as measured using Illumina sequencing and MALDI-TOF
Environ Res 140255ndash267
Martin M 2011 Cutadapt removes adapter sequences from high-
throughput sequencing reads EMBnet J 17(1)10ndash12
Marx CJ et al 2012 Complete genome sequences of six strains of the
genus Methylobacterium b J Bacteriol 194(17)4746
Noguchi T Hayashi H Shimada K Takaichi S Tasumi M 1992 In vivo
states and functions of carotenoids in an aerobic photosynthetic bac-
terium Erythrobacter longus Photosynth Res 31(1)21ndash30
Olm MR Brown CT Brooks B Banfield JF 2017 dRep a tool for fast and
accurate genomic comparisons that enables improved genome recov-
ery from metagenomes through de-replication ISME J
11(12)2864ndash2868
Overbeek R et al 2014 The SEED and the Rapid Annotation of microbial
genomes using Subsystems Technology (RAST) Nucl Acids Res
42(D1)D206ndashD214
Parks DH Imelfort M Skennerton CT Hugenholtz P Tyson GW 2015
CheckM assessing the quality of microbial genomes recovered from
isolates single cells and metagenomes Genome Res 25(7)1043ndash1055
Remus-Emsermann MNP et al 2014 Spatial distribution analyses of nat-
ural phyllosphere-colonizing bacteria on Arabidopsis thaliana revealed
by fluorescence in situ hybridization Environ Microbiol
16(7)2329ndash2340
Stamatakis A 2006 RAxML-VI-HPC maximum likelihood-based phyloge-
netic analyses with thousands of taxa and mixed models
Bioinformatics 22(21)2688ndash2690
UN 2017 World population prospects the 2017 revision Data Booklet
httpspopulationunorgwppPublicationsFilesWPP2019_DataBook-
letpdf
Vesth T Lagesen K Acar euroO Ussery D 2013 CMG-biotools a free work-
bench for basic comparative microbial genomics PLoS One
8(4)e60120
Vorholt JA 2012 Microbial life in the phyllosphere Nat Rev Microbiol
10(12)828ndash840
Wick RR Judd LM Gorrie CL Holt KE 2017 Unicycler resolving bacterial
genome assemblies from short and long sequencing reads PLoS
Comput Biol 13(6)e1005595
Wilson M Lindow SE 1994a Coexistence among epiphytic bacterial pop-
ulations mediated through nutritional resource partitioning Appl
Environ Microbiol 604468ndash77
Wilson M Lindow SE 1994b Ecological similarity and coexistence of epi-
phytic ice-nucleating (ice) pseudomonas syringae strains and a non-
ice-nucleating (ice) biological control agent Appl Environ Microbiol
60(9)3128ndash3137
Woodward FI Lomas MR 2004 Vegetation dynamics ndash simulating
responses to climatic change Biol Rev 79(3)643ndash670
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2907
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Xu J Gordon JI 2003 Honor thy symbionts Proc Natl Acad Sci USA
100(18)10452ndash10459
Yurkov V Csotonyi JT 2009 New light on aerobic anoxygenic photo-
trophs In Hunter NC Daldal F Thurnauer MC Beatty JT editors
The purple photosynthetic bacteria Dordrecht Springer thorn Business
Media BV p 31ndash55 doi101007978-1-4020-8815-5_3
Yurkov V Hughes E 2017 Aerobic Anoxygenic Phototrophs four
decades of mystery In Hallenbeck PC editor Modern topics in the
phototrophic prokaryotes environmental and applied aspects p
193ndash214
Yutin N et al 2009 BchY-based degenerate primers target all types of
anoxygenic photosynthetic bacteria in a single PCR Appl Environ
Microbiol 75(23)7556ndash7559
Zeng Y Feng F Medova H Dean J Koblizek M 2014 Functional
type 2 photosynthetic reaction centers found in the rare bacterial phy-
lum Gemmatimonadetes Proc Natl Acad Sci USA 111(21)7795ndash7800
Zheng Q et al 2011 Diverse arrangement of photosynthetic gene
clusters in aerobic anoxygenic phototrophic bacteria PLoS One
6(9)e25050
Associate editor Esperanza Martinez-Romero
Zervas et al GBE
2908 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
- evz204-TF1
- evz204-TF2
-
of these two groups rely on the zeaxanthin pathway to pro-
duce carotenoid pigments and most likely nostoxanthin and
erythroxanthin which however do not participate in the
light-harvesting process but rather have a photoprotection
role (Noguchi et al 1992 Yurkov and Csotonyi 2009)
Phylogenetic Observations
In our study we sequenced and analyzed 21 aerobic anoxy-
genic photosynthetic bacteria most of which belonged to the
genus Methylobacterium the presence of which in the
environment has already been discussed previously
Methylobacterium spp are facultative methylotrophs and
can utilize a variety of C1 C2 C3 and C4 compounds
Such compounds are often found in the phyllosphere due
to plant metabolism (Vorholt 2012 Remus-Emsermann
et al 2014) Methylobacterium spp show a wide range in
chromosome size and number of plasmids they contain
(Marx et al 2012) The genus is represented well in
GenBank with 100 complete or draft genomes (last accessed
April 2019) However most of the sequences do not have
FIG 5mdashArchitecture of the different photosynthetic gene clusters Arrows indicate the orientation of the genes Sizes are not representatives of gene
length Only the genes shared by all PGCs were shown Genes that may be missing from some strains are denoted with parentheses on the figurersquos legend
Table 3
General Descriptive Statistics of the Four Complete AAP Bacterial Genomes
Taxon Genome Size (bp) GC Number of Genes Plasmids Plasmids Size (bp) GC
Rhizobium sp WL3 4568855 6160 4450 1 700631 586
mdash mdash mdash 2 85350 575
Roseomonas sp WL45 4533887 708 5042 1 262815 655
1011854 707 1138 2 240556 662
mdash mdash mdash 3 152922 658
mdash mdash mdash 4 85279 646
mdash mdash mdash 5 84774 661
mdash mdash mdash 6 83273 656
mdash mdash mdash 7 65860 648
Alsobacter sp WL4 5405668 679 5013 1 27178 620
mdash mdash mdash 2 9451 606
Methylobacterium sp WL1 6215463 691 6431 1 35731 638
Zervas et al GBE
2904 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
species affiliation (fig 4) Thus we were not able to assign any
of the 18 isolates to specific Methylobacterium species apart
perhaps from group 2 (WL1 WL2 WL7 WL18 WL64) which
is placed as sister to M brachiatum in the MASH-ANI analysis
Even in this case though M pseudosasicola BL36 is also
placed in the same clade as M brachiatum which raises the
question of proper nomenclature This is further evidenced in
the same tree (cyanide clade below) where seven species
share gt90 of their genome For the remaining isolates
WL19 is closely affiliated to other Methylobacterium strains
that have not been identified A similar picture is shown for
WL69 On the other hand WL19 appears to be a novel spe-
cies with unique genomic features and so do the strains from
groups 3 and 4 (red clade) Thus despite how ubiquitous
Methylobacterium is in nature it is clear that part of its diver-
gence is still missing Moreover a critical evaluation of the
current nomenclature of the genus would be beneficial based
on the whole genome sequences because closely related
genomes (MASH-ANI identitygt 97) have been given differ-
ent species affiliations
Apart from this genus we also identified a Rhizobium sp
a Roseomonas sp and a novel Alsobacter strain (WL4) that
belongs in the Methylocystaceae a family of methanotroph
bacteria This is the first time that AAP bacteria from this
family have been isolated and their complete genomes assem-
bled Its closely related Alsobacter spp have been shown to
exhibit high Thallium tolerance (Bao et al 2006) and it will be
interesting to identify heavy-metal resistance genes in non-
heavy-metal polluted phyllosphere samples It is worth men-
tioning that if WL4 contains heavy-metal resistance genes on
top of the PGC this will have been shown for the first time
and its potential to bioremediate heavy-metal contaminated
areas if applied as plant growth promotion and environmental
remediation agent is of high value and needs to further be
explored
Molecular Evolution of the PGCs
For all gene trees of the PGC we observed that Rhizobium sp
WL3 evolves faster than the other isolates followed by
Table 4
Substitutions Per Site in the Different Phylogenetic Analyses
Strain 16S acsF bchL bchY crtB pufL pufMsuper-matrix
CheckM
Roseomonas sp WL45 0096 0147 0100 0235 0250 0095 0132 0160 0184
Rhizobium sp WL3 0184 0622 0767 0690 1370 0343 0525 0580 0372
Alsobacter sp WL4 0151 0462 0233 0490 0870 0258 0334 0380 0403
Methylobacterium sp WL9 0226 0418 0372 0420 0670 0270 0319 0364 0477
Methylobacterium sp WL19 0221 0413 0300 0460 0600 0233 0346 0364 0468
Methylobacterium sp WL69 0231 0373 0333 0470 0620 0260 0358 0368 0468
Methylobacterium sp WL1 0233 0422 0344 0485 0480 0280 0385 0380 0459
Methylobacterium sp WL2 0233 0422 0344 0485 0480 0280 0385 0380 0459
Methylobacterium sp WL7 0238 0427 0344 0480 0480 0278 0385 0380 0459
Methylobacterium sp WL18 0238 0427 0344 0480 0480 0280 0385 0380 0459
Methylobacterium sp WL64 0233 0422 0344 0480 0485 0280 0385 0380 0459
Methylobacterium sp WL6 0199 0378 0256 0425 0680 0208 0311 0340 0455
Methylobacterium sp WL30 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL93 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL116 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL119 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL8 0188 0373 0258 0430 0680 0230 0358 0348 0459
Methylobacterium sp WL12 0188 0373 0256 0430 0680 0230 0358 0348 0459
Methylobacterium sp WL103 0188 0373 0256 0430 0850 0230 0358 0364 0459
Methylobacterium sp WL120 0188 0373 0258 0430 0680 0233 0358 0348 0459
Methylobacterium sp WL122 0188 0373 0256 0430 0680 0230 0358 0348 0459
Susbstitutions per site
Gro
up1
Gro
up 2
Gro
up 3
Gro
up 4
NOTEmdashAll values correspond to distances from the root where Roseomonas sp WL45 was chosen as outgroup for all the alignments
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2905
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Alsobacter sp WL4 Both these isolates also show much
higher substitution rates compared with their respective 16S
genes and core proteomes (checkM) which indicates that the
PGCs are under relaxed selection pressure In all instances the
phylogenetic trees constructed showed the same topology
and the same grouping of the Methylobacterium isolates
which suggests that the PGCs were incorporated in the
genomes of these strains relatively early in their evolution If
the PGC was free to move laterally then we would have ob-
served very different phylogenetic trees for its genes com-
pared with the 16S and core proteome analyses
Interestingly these observations do not reflect on the archi-
tecture of the PGC where Roseomonas sp WL45 and
Rhizobium sp WL3 contain the PGC in one continuous re-
gion whereas Alsobacter sp WL4 and Methylobacterium sp
WL1 have it in two pieces
For the 18 Methylobacterium isolates group 1 and 2
strains (WL9 WL19 WL69 and WL1 WL2 WL7 WL18
WL64) evolve consistently at different rates compared
with the groups 3 and 4 It appears that there is relatively
strong selection pressure on bchL pufL and pufM which
in all cases are more similar to each other and also show
slower molecular evolution rates compared with ascF
bchY and crtB which in all Methylobacterium isolates
evolve at least twice as fast compared with the 16S
gene (table 4) They also show fewer substitutions per
site compared with the core proteome analysis This result
partially contrasted with previously adopted methods of
using bchY as a marker gene for RC1 clusters (eg
Atamna-Ismaeel and Finkel 2012a) On the other hand
the use of pufM appears more sensible It would how-
ever be more suitable to investigate the possibility of us-
ing other genes for metagenome amplicon sequencing
approaches such as pufL or bchL which show a smaller
degree of divergence in different genera thus universal
primers might be more successful in detecting a wide va-
riety of AAP bacteria in environmental samples
Interestingly while Groups 1 and 2 isolates exhibit higher
substitution rates of the PGC genes compared with
Groups 3 and 4 this is not the case for crtB which evolves
a lot slower and also pufM which shows similar molec-
ular evolution in the genus Overall it seems that the
PGCs were incorporated in ancestral strains that later di-
verged to give rise to the different AAP taxa This is fur-
ther corroborated by the fact that almost all phylogenetic
trees (eg fig 5) are consistent and in agreement with
both the 16S phylogenetic tree (fig 1) as well as the phy-
logenomic tree (fig 2) If the genes of the PGCs had been
transferred to these AAP several times there would have
been significant differences both in the sequence level
and the phylogenetic analysis of the different genes
which would be expected to result in different topologies
and diverse distances from the root even for closelymdash
based on 16S analysismdashrelated strains
The Significance of AAP Bacteria in Wheat Phyllosphere
In this study we isolated for the first time a variety of AAP
bacteria that are present in the wheat phyllosphere adding to
the metagenomics knowledge provided earlier from five other
plants (Atamna-Ismaeel and Finkel 2012a 2012b) The phyllo-
sphere is a rather harsh environment with little water avail-
ability scarce resources very few suitable locations extreme
temperature difference between day and night and high
dosages of UV radiation (Vorholt 2012) AAP bacteria have
the ability to utilize light in photophosphorylation processes
through which they produce ATP and at the same time funnel
the few available nutrients in other metabolic pathways
(Yurkov and Hughes 2017) This has been shown to give
them an edge in vitro after exposure to light compared with
nonAAP bacteria (Ferrera et al 2011) Considering that phyl-
losphere bacteria protect the plant from pathogenic microbes
by secreting metabolites and by occupying the available
niches being able to persevere and outgrow other bacteria
gives a significant edge to AAP taxa and probably points
toward the suggestion of positive selection of such bacteria
in their phyllosphere by the plants themselves as they appear
to be absent from the soil (Atamna-Ismaeel and Finkel
2012b) At this point it is not clear exactly how AAP bacteria
benefit the plant especially since the phyllosphere is a com-
plex habitat with an ever-changing harsh environment In the
past years great efforts have been carried out on how to
improve crop production by studying the rhizosphere It has
also been suggested that the phyllosphere communities play a
significant role in plant health and growth which ultimately
leads to increased yields (Lindow and Brandl 2003) Given the
current projections of world population growth and the sup-
ply of food (UN 2017) it becomes evident that more in-depth
studies investigating the phyllosphere of plants and especially
crops and identifying the underlying drivers of the bacterial
communities and their interactions with their plant hosts is of
paramount importance Thus by isolating maintaining and
analyzing the genomes of phyllosphere isolates we may iden-
tify strains that may prove useful as plant growth promoting
agents in future field trials investigating the effect of phyllo-
sphere inoculation on plant growth health and yield
Supplementary Material
Supplementary data are available at Genome Biology and
Evolution online
Acknowledgments
The authors would like to thank Tue Sparholt Jorgensen for
collecting the wheat sample The authors also want to thank
Tina Thane (AU) Tanja Begovic (AU) and Margit Frederiksen
(NRCWE) for capable laboratory assistance We thank Michal
Koblızek for providing the initial model of the colony infrared
imaging system This work was funded by a Villum
Zervas et al GBE
2906 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Experiment grant (no 17601) and a Marie Skłodowska-Curie
AIAS-COFUND fellowship (EU-FP7 program Grant
Agreement No 609033) awarded to YZ
Author Contributions
AZ YZ LHH designed the study YZ isolated the strains
performed colony infrared imaging and extracted DNA for
Illumina sequencing YZ and AMM performed the MALDI-
TOF MS analysis AZ extracted DNA and performed ONT
Sequencing AZ carried out the bioinformatics and phyloge-
netic analyses AZ and YZ drafted the manuscript AZ
YZ AMM and LHH revised the manuscript for
intellectual content All authors read and approved the
manuscript
Literature Cited
Atamna-Ismaeel N Finkel O 2012a Bacterial anoxygenic photosynthesis
on plant leaf surfaces Environ Microbiol Rep 4(2)209ndash216
Atamna-Ismaeel N Finkel OM 2012b Microbial rhodopsins on leaf sur-
faces of terrestrial plants Environ Microbiol 14(1)140ndash146
Aziz RK et al 2008 The RAST Server rapid annotations using subsystems
technology BMC Genomics 9(1)75
Bankevich A et al 2012 SPAdes a new genome assembly algorithm and
its applications to single-cell sequencing J Comput Biol
19(5)455ndash477
Bao Z Sato Y Kubota M Ohta H 2006 Isolation and characterization of
thallium-tolerant bacteria from heavy metal-polluted river sediment
and non-polluted soils Microb Environ 21(4)251ndash260
Beatie GA 2002 Leaf surface waxes and the process of leaf colonization
by microorganisms In Lindow SE Hecht-Poinar EI Elliott VJ editors
Phyllosphere Microbiology St Paul (USA) APS Press pp 3ndash26
Beattie GA Lindow SE 1999 Bacterial colonization of leaves a spectrum
of strategies Phytopathology 89(5)353ndash359
Brettin T et al 2015 RASTtk a modular and extensible implementation of
the RAST algorithm for building custom annotation pipelines and an-
notating batches of genomes Sci Rep 58365
Carvalho SD Castillo JA 2018 Influence of light on plantndashphyllosphere
interaction Front Plant Sci 91482
Ferrera I Gasol JM Sebastian M Hojerova E Koblızek M 2011
Comparison of growth rates of aerobic anoxygenic phototrophic bac-
teria and other bacterioplankton groups in coastal Mediterranean wa-
ters Appl Environ Microbiol 77(21)7451ndash7458
Ferrera I Sanchez O Kolarova E Koblızek M Gasol JM 2017 Light
enhances the growth rates of natural populations of aerobic anoxy-
genic phototrophic bacteria ISME J 11(10)2391ndash2393
Gorlach K Shingaki R Morisaki H Hattori T 1994 Construction of eco-
collection of paddy field soil bacteria for population analysis J Gen
Appl Microbiol 40(6)509ndash517
Gourion B Rossignol M Vorholt JA Lindow SE 2006 A proteomic study
of Methylobacterium extorquens reveals a response regulator essential
for epiphytic growth wwwintegratedgenomicscom Accessed
November 26 2018
Green PN 2006 Methylobacterium In Dworkin M Falkow S Rosenberg
E Schleifer K Stackebrandt E editors The prokaryotes ndash a handbook
on the biology of bacteria New York Springer p 257ndash265
Hauruseu D Koblızek M 2012 Influence of light on carbon utilization in
aerobic anoxygenic phototrophs Appl Environ Microbiol
78(20)7414ndash7419
Katoh K Misawa K Kuma K Miyata T 2002 MAFFT a novel method for
rapid multiple sequence alignment based on fast Fourier transform
Nucleic Acids Res 30(14)3059ndash3066
Knief C et al 2012 Metaproteogenomic analysis of microbial communi-
ties in the phyllosphere and rhizosphere of rice ISME J
6(7)1378ndash1390
Knief C Frances L Vorholt JA 2010 Competitiveness of diverse methyl-
obacterium strains in the phyllosphere of arabidopsis thaliana and
identification of representative models including M extorquens
PA1 Microb Ecol 60(2)440ndash452
Kwak MJ et al 2014 Genome information of Methylobacterium oryzae a
plantndashprobiotic methylotroph in the phyllosphere PLoS One
9(9)e106704
Lindow SE Brandl MT 2003 Microbiology of the phyllosphere Appl
Environ Microbiol 69(4)1875ndash1883
Madsen AM Zervas A Tendal K Nielsen JL 2015 Microbial diversity in
bioaerosol samples causing ODTS compared to reference bioaerosol
samples as measured using Illumina sequencing and MALDI-TOF
Environ Res 140255ndash267
Martin M 2011 Cutadapt removes adapter sequences from high-
throughput sequencing reads EMBnet J 17(1)10ndash12
Marx CJ et al 2012 Complete genome sequences of six strains of the
genus Methylobacterium b J Bacteriol 194(17)4746
Noguchi T Hayashi H Shimada K Takaichi S Tasumi M 1992 In vivo
states and functions of carotenoids in an aerobic photosynthetic bac-
terium Erythrobacter longus Photosynth Res 31(1)21ndash30
Olm MR Brown CT Brooks B Banfield JF 2017 dRep a tool for fast and
accurate genomic comparisons that enables improved genome recov-
ery from metagenomes through de-replication ISME J
11(12)2864ndash2868
Overbeek R et al 2014 The SEED and the Rapid Annotation of microbial
genomes using Subsystems Technology (RAST) Nucl Acids Res
42(D1)D206ndashD214
Parks DH Imelfort M Skennerton CT Hugenholtz P Tyson GW 2015
CheckM assessing the quality of microbial genomes recovered from
isolates single cells and metagenomes Genome Res 25(7)1043ndash1055
Remus-Emsermann MNP et al 2014 Spatial distribution analyses of nat-
ural phyllosphere-colonizing bacteria on Arabidopsis thaliana revealed
by fluorescence in situ hybridization Environ Microbiol
16(7)2329ndash2340
Stamatakis A 2006 RAxML-VI-HPC maximum likelihood-based phyloge-
netic analyses with thousands of taxa and mixed models
Bioinformatics 22(21)2688ndash2690
UN 2017 World population prospects the 2017 revision Data Booklet
httpspopulationunorgwppPublicationsFilesWPP2019_DataBook-
letpdf
Vesth T Lagesen K Acar euroO Ussery D 2013 CMG-biotools a free work-
bench for basic comparative microbial genomics PLoS One
8(4)e60120
Vorholt JA 2012 Microbial life in the phyllosphere Nat Rev Microbiol
10(12)828ndash840
Wick RR Judd LM Gorrie CL Holt KE 2017 Unicycler resolving bacterial
genome assemblies from short and long sequencing reads PLoS
Comput Biol 13(6)e1005595
Wilson M Lindow SE 1994a Coexistence among epiphytic bacterial pop-
ulations mediated through nutritional resource partitioning Appl
Environ Microbiol 604468ndash77
Wilson M Lindow SE 1994b Ecological similarity and coexistence of epi-
phytic ice-nucleating (ice) pseudomonas syringae strains and a non-
ice-nucleating (ice) biological control agent Appl Environ Microbiol
60(9)3128ndash3137
Woodward FI Lomas MR 2004 Vegetation dynamics ndash simulating
responses to climatic change Biol Rev 79(3)643ndash670
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2907
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Xu J Gordon JI 2003 Honor thy symbionts Proc Natl Acad Sci USA
100(18)10452ndash10459
Yurkov V Csotonyi JT 2009 New light on aerobic anoxygenic photo-
trophs In Hunter NC Daldal F Thurnauer MC Beatty JT editors
The purple photosynthetic bacteria Dordrecht Springer thorn Business
Media BV p 31ndash55 doi101007978-1-4020-8815-5_3
Yurkov V Hughes E 2017 Aerobic Anoxygenic Phototrophs four
decades of mystery In Hallenbeck PC editor Modern topics in the
phototrophic prokaryotes environmental and applied aspects p
193ndash214
Yutin N et al 2009 BchY-based degenerate primers target all types of
anoxygenic photosynthetic bacteria in a single PCR Appl Environ
Microbiol 75(23)7556ndash7559
Zeng Y Feng F Medova H Dean J Koblizek M 2014 Functional
type 2 photosynthetic reaction centers found in the rare bacterial phy-
lum Gemmatimonadetes Proc Natl Acad Sci USA 111(21)7795ndash7800
Zheng Q et al 2011 Diverse arrangement of photosynthetic gene
clusters in aerobic anoxygenic phototrophic bacteria PLoS One
6(9)e25050
Associate editor Esperanza Martinez-Romero
Zervas et al GBE
2908 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
- evz204-TF1
- evz204-TF2
-
species affiliation (fig 4) Thus we were not able to assign any
of the 18 isolates to specific Methylobacterium species apart
perhaps from group 2 (WL1 WL2 WL7 WL18 WL64) which
is placed as sister to M brachiatum in the MASH-ANI analysis
Even in this case though M pseudosasicola BL36 is also
placed in the same clade as M brachiatum which raises the
question of proper nomenclature This is further evidenced in
the same tree (cyanide clade below) where seven species
share gt90 of their genome For the remaining isolates
WL19 is closely affiliated to other Methylobacterium strains
that have not been identified A similar picture is shown for
WL69 On the other hand WL19 appears to be a novel spe-
cies with unique genomic features and so do the strains from
groups 3 and 4 (red clade) Thus despite how ubiquitous
Methylobacterium is in nature it is clear that part of its diver-
gence is still missing Moreover a critical evaluation of the
current nomenclature of the genus would be beneficial based
on the whole genome sequences because closely related
genomes (MASH-ANI identitygt 97) have been given differ-
ent species affiliations
Apart from this genus we also identified a Rhizobium sp
a Roseomonas sp and a novel Alsobacter strain (WL4) that
belongs in the Methylocystaceae a family of methanotroph
bacteria This is the first time that AAP bacteria from this
family have been isolated and their complete genomes assem-
bled Its closely related Alsobacter spp have been shown to
exhibit high Thallium tolerance (Bao et al 2006) and it will be
interesting to identify heavy-metal resistance genes in non-
heavy-metal polluted phyllosphere samples It is worth men-
tioning that if WL4 contains heavy-metal resistance genes on
top of the PGC this will have been shown for the first time
and its potential to bioremediate heavy-metal contaminated
areas if applied as plant growth promotion and environmental
remediation agent is of high value and needs to further be
explored
Molecular Evolution of the PGCs
For all gene trees of the PGC we observed that Rhizobium sp
WL3 evolves faster than the other isolates followed by
Table 4
Substitutions Per Site in the Different Phylogenetic Analyses
Strain 16S acsF bchL bchY crtB pufL pufMsuper-matrix
CheckM
Roseomonas sp WL45 0096 0147 0100 0235 0250 0095 0132 0160 0184
Rhizobium sp WL3 0184 0622 0767 0690 1370 0343 0525 0580 0372
Alsobacter sp WL4 0151 0462 0233 0490 0870 0258 0334 0380 0403
Methylobacterium sp WL9 0226 0418 0372 0420 0670 0270 0319 0364 0477
Methylobacterium sp WL19 0221 0413 0300 0460 0600 0233 0346 0364 0468
Methylobacterium sp WL69 0231 0373 0333 0470 0620 0260 0358 0368 0468
Methylobacterium sp WL1 0233 0422 0344 0485 0480 0280 0385 0380 0459
Methylobacterium sp WL2 0233 0422 0344 0485 0480 0280 0385 0380 0459
Methylobacterium sp WL7 0238 0427 0344 0480 0480 0278 0385 0380 0459
Methylobacterium sp WL18 0238 0427 0344 0480 0480 0280 0385 0380 0459
Methylobacterium sp WL64 0233 0422 0344 0480 0485 0280 0385 0380 0459
Methylobacterium sp WL6 0199 0378 0256 0425 0680 0208 0311 0340 0455
Methylobacterium sp WL30 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL93 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL116 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL119 0199 0373 0256 0430 0680 0210 0334 0340 0455
Methylobacterium sp WL8 0188 0373 0258 0430 0680 0230 0358 0348 0459
Methylobacterium sp WL12 0188 0373 0256 0430 0680 0230 0358 0348 0459
Methylobacterium sp WL103 0188 0373 0256 0430 0850 0230 0358 0364 0459
Methylobacterium sp WL120 0188 0373 0258 0430 0680 0233 0358 0348 0459
Methylobacterium sp WL122 0188 0373 0256 0430 0680 0230 0358 0348 0459
Susbstitutions per site
Gro
up1
Gro
up 2
Gro
up 3
Gro
up 4
NOTEmdashAll values correspond to distances from the root where Roseomonas sp WL45 was chosen as outgroup for all the alignments
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2905
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Alsobacter sp WL4 Both these isolates also show much
higher substitution rates compared with their respective 16S
genes and core proteomes (checkM) which indicates that the
PGCs are under relaxed selection pressure In all instances the
phylogenetic trees constructed showed the same topology
and the same grouping of the Methylobacterium isolates
which suggests that the PGCs were incorporated in the
genomes of these strains relatively early in their evolution If
the PGC was free to move laterally then we would have ob-
served very different phylogenetic trees for its genes com-
pared with the 16S and core proteome analyses
Interestingly these observations do not reflect on the archi-
tecture of the PGC where Roseomonas sp WL45 and
Rhizobium sp WL3 contain the PGC in one continuous re-
gion whereas Alsobacter sp WL4 and Methylobacterium sp
WL1 have it in two pieces
For the 18 Methylobacterium isolates group 1 and 2
strains (WL9 WL19 WL69 and WL1 WL2 WL7 WL18
WL64) evolve consistently at different rates compared
with the groups 3 and 4 It appears that there is relatively
strong selection pressure on bchL pufL and pufM which
in all cases are more similar to each other and also show
slower molecular evolution rates compared with ascF
bchY and crtB which in all Methylobacterium isolates
evolve at least twice as fast compared with the 16S
gene (table 4) They also show fewer substitutions per
site compared with the core proteome analysis This result
partially contrasted with previously adopted methods of
using bchY as a marker gene for RC1 clusters (eg
Atamna-Ismaeel and Finkel 2012a) On the other hand
the use of pufM appears more sensible It would how-
ever be more suitable to investigate the possibility of us-
ing other genes for metagenome amplicon sequencing
approaches such as pufL or bchL which show a smaller
degree of divergence in different genera thus universal
primers might be more successful in detecting a wide va-
riety of AAP bacteria in environmental samples
Interestingly while Groups 1 and 2 isolates exhibit higher
substitution rates of the PGC genes compared with
Groups 3 and 4 this is not the case for crtB which evolves
a lot slower and also pufM which shows similar molec-
ular evolution in the genus Overall it seems that the
PGCs were incorporated in ancestral strains that later di-
verged to give rise to the different AAP taxa This is fur-
ther corroborated by the fact that almost all phylogenetic
trees (eg fig 5) are consistent and in agreement with
both the 16S phylogenetic tree (fig 1) as well as the phy-
logenomic tree (fig 2) If the genes of the PGCs had been
transferred to these AAP several times there would have
been significant differences both in the sequence level
and the phylogenetic analysis of the different genes
which would be expected to result in different topologies
and diverse distances from the root even for closelymdash
based on 16S analysismdashrelated strains
The Significance of AAP Bacteria in Wheat Phyllosphere
In this study we isolated for the first time a variety of AAP
bacteria that are present in the wheat phyllosphere adding to
the metagenomics knowledge provided earlier from five other
plants (Atamna-Ismaeel and Finkel 2012a 2012b) The phyllo-
sphere is a rather harsh environment with little water avail-
ability scarce resources very few suitable locations extreme
temperature difference between day and night and high
dosages of UV radiation (Vorholt 2012) AAP bacteria have
the ability to utilize light in photophosphorylation processes
through which they produce ATP and at the same time funnel
the few available nutrients in other metabolic pathways
(Yurkov and Hughes 2017) This has been shown to give
them an edge in vitro after exposure to light compared with
nonAAP bacteria (Ferrera et al 2011) Considering that phyl-
losphere bacteria protect the plant from pathogenic microbes
by secreting metabolites and by occupying the available
niches being able to persevere and outgrow other bacteria
gives a significant edge to AAP taxa and probably points
toward the suggestion of positive selection of such bacteria
in their phyllosphere by the plants themselves as they appear
to be absent from the soil (Atamna-Ismaeel and Finkel
2012b) At this point it is not clear exactly how AAP bacteria
benefit the plant especially since the phyllosphere is a com-
plex habitat with an ever-changing harsh environment In the
past years great efforts have been carried out on how to
improve crop production by studying the rhizosphere It has
also been suggested that the phyllosphere communities play a
significant role in plant health and growth which ultimately
leads to increased yields (Lindow and Brandl 2003) Given the
current projections of world population growth and the sup-
ply of food (UN 2017) it becomes evident that more in-depth
studies investigating the phyllosphere of plants and especially
crops and identifying the underlying drivers of the bacterial
communities and their interactions with their plant hosts is of
paramount importance Thus by isolating maintaining and
analyzing the genomes of phyllosphere isolates we may iden-
tify strains that may prove useful as plant growth promoting
agents in future field trials investigating the effect of phyllo-
sphere inoculation on plant growth health and yield
Supplementary Material
Supplementary data are available at Genome Biology and
Evolution online
Acknowledgments
The authors would like to thank Tue Sparholt Jorgensen for
collecting the wheat sample The authors also want to thank
Tina Thane (AU) Tanja Begovic (AU) and Margit Frederiksen
(NRCWE) for capable laboratory assistance We thank Michal
Koblızek for providing the initial model of the colony infrared
imaging system This work was funded by a Villum
Zervas et al GBE
2906 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Experiment grant (no 17601) and a Marie Skłodowska-Curie
AIAS-COFUND fellowship (EU-FP7 program Grant
Agreement No 609033) awarded to YZ
Author Contributions
AZ YZ LHH designed the study YZ isolated the strains
performed colony infrared imaging and extracted DNA for
Illumina sequencing YZ and AMM performed the MALDI-
TOF MS analysis AZ extracted DNA and performed ONT
Sequencing AZ carried out the bioinformatics and phyloge-
netic analyses AZ and YZ drafted the manuscript AZ
YZ AMM and LHH revised the manuscript for
intellectual content All authors read and approved the
manuscript
Literature Cited
Atamna-Ismaeel N Finkel O 2012a Bacterial anoxygenic photosynthesis
on plant leaf surfaces Environ Microbiol Rep 4(2)209ndash216
Atamna-Ismaeel N Finkel OM 2012b Microbial rhodopsins on leaf sur-
faces of terrestrial plants Environ Microbiol 14(1)140ndash146
Aziz RK et al 2008 The RAST Server rapid annotations using subsystems
technology BMC Genomics 9(1)75
Bankevich A et al 2012 SPAdes a new genome assembly algorithm and
its applications to single-cell sequencing J Comput Biol
19(5)455ndash477
Bao Z Sato Y Kubota M Ohta H 2006 Isolation and characterization of
thallium-tolerant bacteria from heavy metal-polluted river sediment
and non-polluted soils Microb Environ 21(4)251ndash260
Beatie GA 2002 Leaf surface waxes and the process of leaf colonization
by microorganisms In Lindow SE Hecht-Poinar EI Elliott VJ editors
Phyllosphere Microbiology St Paul (USA) APS Press pp 3ndash26
Beattie GA Lindow SE 1999 Bacterial colonization of leaves a spectrum
of strategies Phytopathology 89(5)353ndash359
Brettin T et al 2015 RASTtk a modular and extensible implementation of
the RAST algorithm for building custom annotation pipelines and an-
notating batches of genomes Sci Rep 58365
Carvalho SD Castillo JA 2018 Influence of light on plantndashphyllosphere
interaction Front Plant Sci 91482
Ferrera I Gasol JM Sebastian M Hojerova E Koblızek M 2011
Comparison of growth rates of aerobic anoxygenic phototrophic bac-
teria and other bacterioplankton groups in coastal Mediterranean wa-
ters Appl Environ Microbiol 77(21)7451ndash7458
Ferrera I Sanchez O Kolarova E Koblızek M Gasol JM 2017 Light
enhances the growth rates of natural populations of aerobic anoxy-
genic phototrophic bacteria ISME J 11(10)2391ndash2393
Gorlach K Shingaki R Morisaki H Hattori T 1994 Construction of eco-
collection of paddy field soil bacteria for population analysis J Gen
Appl Microbiol 40(6)509ndash517
Gourion B Rossignol M Vorholt JA Lindow SE 2006 A proteomic study
of Methylobacterium extorquens reveals a response regulator essential
for epiphytic growth wwwintegratedgenomicscom Accessed
November 26 2018
Green PN 2006 Methylobacterium In Dworkin M Falkow S Rosenberg
E Schleifer K Stackebrandt E editors The prokaryotes ndash a handbook
on the biology of bacteria New York Springer p 257ndash265
Hauruseu D Koblızek M 2012 Influence of light on carbon utilization in
aerobic anoxygenic phototrophs Appl Environ Microbiol
78(20)7414ndash7419
Katoh K Misawa K Kuma K Miyata T 2002 MAFFT a novel method for
rapid multiple sequence alignment based on fast Fourier transform
Nucleic Acids Res 30(14)3059ndash3066
Knief C et al 2012 Metaproteogenomic analysis of microbial communi-
ties in the phyllosphere and rhizosphere of rice ISME J
6(7)1378ndash1390
Knief C Frances L Vorholt JA 2010 Competitiveness of diverse methyl-
obacterium strains in the phyllosphere of arabidopsis thaliana and
identification of representative models including M extorquens
PA1 Microb Ecol 60(2)440ndash452
Kwak MJ et al 2014 Genome information of Methylobacterium oryzae a
plantndashprobiotic methylotroph in the phyllosphere PLoS One
9(9)e106704
Lindow SE Brandl MT 2003 Microbiology of the phyllosphere Appl
Environ Microbiol 69(4)1875ndash1883
Madsen AM Zervas A Tendal K Nielsen JL 2015 Microbial diversity in
bioaerosol samples causing ODTS compared to reference bioaerosol
samples as measured using Illumina sequencing and MALDI-TOF
Environ Res 140255ndash267
Martin M 2011 Cutadapt removes adapter sequences from high-
throughput sequencing reads EMBnet J 17(1)10ndash12
Marx CJ et al 2012 Complete genome sequences of six strains of the
genus Methylobacterium b J Bacteriol 194(17)4746
Noguchi T Hayashi H Shimada K Takaichi S Tasumi M 1992 In vivo
states and functions of carotenoids in an aerobic photosynthetic bac-
terium Erythrobacter longus Photosynth Res 31(1)21ndash30
Olm MR Brown CT Brooks B Banfield JF 2017 dRep a tool for fast and
accurate genomic comparisons that enables improved genome recov-
ery from metagenomes through de-replication ISME J
11(12)2864ndash2868
Overbeek R et al 2014 The SEED and the Rapid Annotation of microbial
genomes using Subsystems Technology (RAST) Nucl Acids Res
42(D1)D206ndashD214
Parks DH Imelfort M Skennerton CT Hugenholtz P Tyson GW 2015
CheckM assessing the quality of microbial genomes recovered from
isolates single cells and metagenomes Genome Res 25(7)1043ndash1055
Remus-Emsermann MNP et al 2014 Spatial distribution analyses of nat-
ural phyllosphere-colonizing bacteria on Arabidopsis thaliana revealed
by fluorescence in situ hybridization Environ Microbiol
16(7)2329ndash2340
Stamatakis A 2006 RAxML-VI-HPC maximum likelihood-based phyloge-
netic analyses with thousands of taxa and mixed models
Bioinformatics 22(21)2688ndash2690
UN 2017 World population prospects the 2017 revision Data Booklet
httpspopulationunorgwppPublicationsFilesWPP2019_DataBook-
letpdf
Vesth T Lagesen K Acar euroO Ussery D 2013 CMG-biotools a free work-
bench for basic comparative microbial genomics PLoS One
8(4)e60120
Vorholt JA 2012 Microbial life in the phyllosphere Nat Rev Microbiol
10(12)828ndash840
Wick RR Judd LM Gorrie CL Holt KE 2017 Unicycler resolving bacterial
genome assemblies from short and long sequencing reads PLoS
Comput Biol 13(6)e1005595
Wilson M Lindow SE 1994a Coexistence among epiphytic bacterial pop-
ulations mediated through nutritional resource partitioning Appl
Environ Microbiol 604468ndash77
Wilson M Lindow SE 1994b Ecological similarity and coexistence of epi-
phytic ice-nucleating (ice) pseudomonas syringae strains and a non-
ice-nucleating (ice) biological control agent Appl Environ Microbiol
60(9)3128ndash3137
Woodward FI Lomas MR 2004 Vegetation dynamics ndash simulating
responses to climatic change Biol Rev 79(3)643ndash670
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2907
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Xu J Gordon JI 2003 Honor thy symbionts Proc Natl Acad Sci USA
100(18)10452ndash10459
Yurkov V Csotonyi JT 2009 New light on aerobic anoxygenic photo-
trophs In Hunter NC Daldal F Thurnauer MC Beatty JT editors
The purple photosynthetic bacteria Dordrecht Springer thorn Business
Media BV p 31ndash55 doi101007978-1-4020-8815-5_3
Yurkov V Hughes E 2017 Aerobic Anoxygenic Phototrophs four
decades of mystery In Hallenbeck PC editor Modern topics in the
phototrophic prokaryotes environmental and applied aspects p
193ndash214
Yutin N et al 2009 BchY-based degenerate primers target all types of
anoxygenic photosynthetic bacteria in a single PCR Appl Environ
Microbiol 75(23)7556ndash7559
Zeng Y Feng F Medova H Dean J Koblizek M 2014 Functional
type 2 photosynthetic reaction centers found in the rare bacterial phy-
lum Gemmatimonadetes Proc Natl Acad Sci USA 111(21)7795ndash7800
Zheng Q et al 2011 Diverse arrangement of photosynthetic gene
clusters in aerobic anoxygenic phototrophic bacteria PLoS One
6(9)e25050
Associate editor Esperanza Martinez-Romero
Zervas et al GBE
2908 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
- evz204-TF1
- evz204-TF2
-
Alsobacter sp WL4 Both these isolates also show much
higher substitution rates compared with their respective 16S
genes and core proteomes (checkM) which indicates that the
PGCs are under relaxed selection pressure In all instances the
phylogenetic trees constructed showed the same topology
and the same grouping of the Methylobacterium isolates
which suggests that the PGCs were incorporated in the
genomes of these strains relatively early in their evolution If
the PGC was free to move laterally then we would have ob-
served very different phylogenetic trees for its genes com-
pared with the 16S and core proteome analyses
Interestingly these observations do not reflect on the archi-
tecture of the PGC where Roseomonas sp WL45 and
Rhizobium sp WL3 contain the PGC in one continuous re-
gion whereas Alsobacter sp WL4 and Methylobacterium sp
WL1 have it in two pieces
For the 18 Methylobacterium isolates group 1 and 2
strains (WL9 WL19 WL69 and WL1 WL2 WL7 WL18
WL64) evolve consistently at different rates compared
with the groups 3 and 4 It appears that there is relatively
strong selection pressure on bchL pufL and pufM which
in all cases are more similar to each other and also show
slower molecular evolution rates compared with ascF
bchY and crtB which in all Methylobacterium isolates
evolve at least twice as fast compared with the 16S
gene (table 4) They also show fewer substitutions per
site compared with the core proteome analysis This result
partially contrasted with previously adopted methods of
using bchY as a marker gene for RC1 clusters (eg
Atamna-Ismaeel and Finkel 2012a) On the other hand
the use of pufM appears more sensible It would how-
ever be more suitable to investigate the possibility of us-
ing other genes for metagenome amplicon sequencing
approaches such as pufL or bchL which show a smaller
degree of divergence in different genera thus universal
primers might be more successful in detecting a wide va-
riety of AAP bacteria in environmental samples
Interestingly while Groups 1 and 2 isolates exhibit higher
substitution rates of the PGC genes compared with
Groups 3 and 4 this is not the case for crtB which evolves
a lot slower and also pufM which shows similar molec-
ular evolution in the genus Overall it seems that the
PGCs were incorporated in ancestral strains that later di-
verged to give rise to the different AAP taxa This is fur-
ther corroborated by the fact that almost all phylogenetic
trees (eg fig 5) are consistent and in agreement with
both the 16S phylogenetic tree (fig 1) as well as the phy-
logenomic tree (fig 2) If the genes of the PGCs had been
transferred to these AAP several times there would have
been significant differences both in the sequence level
and the phylogenetic analysis of the different genes
which would be expected to result in different topologies
and diverse distances from the root even for closelymdash
based on 16S analysismdashrelated strains
The Significance of AAP Bacteria in Wheat Phyllosphere
In this study we isolated for the first time a variety of AAP
bacteria that are present in the wheat phyllosphere adding to
the metagenomics knowledge provided earlier from five other
plants (Atamna-Ismaeel and Finkel 2012a 2012b) The phyllo-
sphere is a rather harsh environment with little water avail-
ability scarce resources very few suitable locations extreme
temperature difference between day and night and high
dosages of UV radiation (Vorholt 2012) AAP bacteria have
the ability to utilize light in photophosphorylation processes
through which they produce ATP and at the same time funnel
the few available nutrients in other metabolic pathways
(Yurkov and Hughes 2017) This has been shown to give
them an edge in vitro after exposure to light compared with
nonAAP bacteria (Ferrera et al 2011) Considering that phyl-
losphere bacteria protect the plant from pathogenic microbes
by secreting metabolites and by occupying the available
niches being able to persevere and outgrow other bacteria
gives a significant edge to AAP taxa and probably points
toward the suggestion of positive selection of such bacteria
in their phyllosphere by the plants themselves as they appear
to be absent from the soil (Atamna-Ismaeel and Finkel
2012b) At this point it is not clear exactly how AAP bacteria
benefit the plant especially since the phyllosphere is a com-
plex habitat with an ever-changing harsh environment In the
past years great efforts have been carried out on how to
improve crop production by studying the rhizosphere It has
also been suggested that the phyllosphere communities play a
significant role in plant health and growth which ultimately
leads to increased yields (Lindow and Brandl 2003) Given the
current projections of world population growth and the sup-
ply of food (UN 2017) it becomes evident that more in-depth
studies investigating the phyllosphere of plants and especially
crops and identifying the underlying drivers of the bacterial
communities and their interactions with their plant hosts is of
paramount importance Thus by isolating maintaining and
analyzing the genomes of phyllosphere isolates we may iden-
tify strains that may prove useful as plant growth promoting
agents in future field trials investigating the effect of phyllo-
sphere inoculation on plant growth health and yield
Supplementary Material
Supplementary data are available at Genome Biology and
Evolution online
Acknowledgments
The authors would like to thank Tue Sparholt Jorgensen for
collecting the wheat sample The authors also want to thank
Tina Thane (AU) Tanja Begovic (AU) and Margit Frederiksen
(NRCWE) for capable laboratory assistance We thank Michal
Koblızek for providing the initial model of the colony infrared
imaging system This work was funded by a Villum
Zervas et al GBE
2906 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Experiment grant (no 17601) and a Marie Skłodowska-Curie
AIAS-COFUND fellowship (EU-FP7 program Grant
Agreement No 609033) awarded to YZ
Author Contributions
AZ YZ LHH designed the study YZ isolated the strains
performed colony infrared imaging and extracted DNA for
Illumina sequencing YZ and AMM performed the MALDI-
TOF MS analysis AZ extracted DNA and performed ONT
Sequencing AZ carried out the bioinformatics and phyloge-
netic analyses AZ and YZ drafted the manuscript AZ
YZ AMM and LHH revised the manuscript for
intellectual content All authors read and approved the
manuscript
Literature Cited
Atamna-Ismaeel N Finkel O 2012a Bacterial anoxygenic photosynthesis
on plant leaf surfaces Environ Microbiol Rep 4(2)209ndash216
Atamna-Ismaeel N Finkel OM 2012b Microbial rhodopsins on leaf sur-
faces of terrestrial plants Environ Microbiol 14(1)140ndash146
Aziz RK et al 2008 The RAST Server rapid annotations using subsystems
technology BMC Genomics 9(1)75
Bankevich A et al 2012 SPAdes a new genome assembly algorithm and
its applications to single-cell sequencing J Comput Biol
19(5)455ndash477
Bao Z Sato Y Kubota M Ohta H 2006 Isolation and characterization of
thallium-tolerant bacteria from heavy metal-polluted river sediment
and non-polluted soils Microb Environ 21(4)251ndash260
Beatie GA 2002 Leaf surface waxes and the process of leaf colonization
by microorganisms In Lindow SE Hecht-Poinar EI Elliott VJ editors
Phyllosphere Microbiology St Paul (USA) APS Press pp 3ndash26
Beattie GA Lindow SE 1999 Bacterial colonization of leaves a spectrum
of strategies Phytopathology 89(5)353ndash359
Brettin T et al 2015 RASTtk a modular and extensible implementation of
the RAST algorithm for building custom annotation pipelines and an-
notating batches of genomes Sci Rep 58365
Carvalho SD Castillo JA 2018 Influence of light on plantndashphyllosphere
interaction Front Plant Sci 91482
Ferrera I Gasol JM Sebastian M Hojerova E Koblızek M 2011
Comparison of growth rates of aerobic anoxygenic phototrophic bac-
teria and other bacterioplankton groups in coastal Mediterranean wa-
ters Appl Environ Microbiol 77(21)7451ndash7458
Ferrera I Sanchez O Kolarova E Koblızek M Gasol JM 2017 Light
enhances the growth rates of natural populations of aerobic anoxy-
genic phototrophic bacteria ISME J 11(10)2391ndash2393
Gorlach K Shingaki R Morisaki H Hattori T 1994 Construction of eco-
collection of paddy field soil bacteria for population analysis J Gen
Appl Microbiol 40(6)509ndash517
Gourion B Rossignol M Vorholt JA Lindow SE 2006 A proteomic study
of Methylobacterium extorquens reveals a response regulator essential
for epiphytic growth wwwintegratedgenomicscom Accessed
November 26 2018
Green PN 2006 Methylobacterium In Dworkin M Falkow S Rosenberg
E Schleifer K Stackebrandt E editors The prokaryotes ndash a handbook
on the biology of bacteria New York Springer p 257ndash265
Hauruseu D Koblızek M 2012 Influence of light on carbon utilization in
aerobic anoxygenic phototrophs Appl Environ Microbiol
78(20)7414ndash7419
Katoh K Misawa K Kuma K Miyata T 2002 MAFFT a novel method for
rapid multiple sequence alignment based on fast Fourier transform
Nucleic Acids Res 30(14)3059ndash3066
Knief C et al 2012 Metaproteogenomic analysis of microbial communi-
ties in the phyllosphere and rhizosphere of rice ISME J
6(7)1378ndash1390
Knief C Frances L Vorholt JA 2010 Competitiveness of diverse methyl-
obacterium strains in the phyllosphere of arabidopsis thaliana and
identification of representative models including M extorquens
PA1 Microb Ecol 60(2)440ndash452
Kwak MJ et al 2014 Genome information of Methylobacterium oryzae a
plantndashprobiotic methylotroph in the phyllosphere PLoS One
9(9)e106704
Lindow SE Brandl MT 2003 Microbiology of the phyllosphere Appl
Environ Microbiol 69(4)1875ndash1883
Madsen AM Zervas A Tendal K Nielsen JL 2015 Microbial diversity in
bioaerosol samples causing ODTS compared to reference bioaerosol
samples as measured using Illumina sequencing and MALDI-TOF
Environ Res 140255ndash267
Martin M 2011 Cutadapt removes adapter sequences from high-
throughput sequencing reads EMBnet J 17(1)10ndash12
Marx CJ et al 2012 Complete genome sequences of six strains of the
genus Methylobacterium b J Bacteriol 194(17)4746
Noguchi T Hayashi H Shimada K Takaichi S Tasumi M 1992 In vivo
states and functions of carotenoids in an aerobic photosynthetic bac-
terium Erythrobacter longus Photosynth Res 31(1)21ndash30
Olm MR Brown CT Brooks B Banfield JF 2017 dRep a tool for fast and
accurate genomic comparisons that enables improved genome recov-
ery from metagenomes through de-replication ISME J
11(12)2864ndash2868
Overbeek R et al 2014 The SEED and the Rapid Annotation of microbial
genomes using Subsystems Technology (RAST) Nucl Acids Res
42(D1)D206ndashD214
Parks DH Imelfort M Skennerton CT Hugenholtz P Tyson GW 2015
CheckM assessing the quality of microbial genomes recovered from
isolates single cells and metagenomes Genome Res 25(7)1043ndash1055
Remus-Emsermann MNP et al 2014 Spatial distribution analyses of nat-
ural phyllosphere-colonizing bacteria on Arabidopsis thaliana revealed
by fluorescence in situ hybridization Environ Microbiol
16(7)2329ndash2340
Stamatakis A 2006 RAxML-VI-HPC maximum likelihood-based phyloge-
netic analyses with thousands of taxa and mixed models
Bioinformatics 22(21)2688ndash2690
UN 2017 World population prospects the 2017 revision Data Booklet
httpspopulationunorgwppPublicationsFilesWPP2019_DataBook-
letpdf
Vesth T Lagesen K Acar euroO Ussery D 2013 CMG-biotools a free work-
bench for basic comparative microbial genomics PLoS One
8(4)e60120
Vorholt JA 2012 Microbial life in the phyllosphere Nat Rev Microbiol
10(12)828ndash840
Wick RR Judd LM Gorrie CL Holt KE 2017 Unicycler resolving bacterial
genome assemblies from short and long sequencing reads PLoS
Comput Biol 13(6)e1005595
Wilson M Lindow SE 1994a Coexistence among epiphytic bacterial pop-
ulations mediated through nutritional resource partitioning Appl
Environ Microbiol 604468ndash77
Wilson M Lindow SE 1994b Ecological similarity and coexistence of epi-
phytic ice-nucleating (ice) pseudomonas syringae strains and a non-
ice-nucleating (ice) biological control agent Appl Environ Microbiol
60(9)3128ndash3137
Woodward FI Lomas MR 2004 Vegetation dynamics ndash simulating
responses to climatic change Biol Rev 79(3)643ndash670
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2907
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Xu J Gordon JI 2003 Honor thy symbionts Proc Natl Acad Sci USA
100(18)10452ndash10459
Yurkov V Csotonyi JT 2009 New light on aerobic anoxygenic photo-
trophs In Hunter NC Daldal F Thurnauer MC Beatty JT editors
The purple photosynthetic bacteria Dordrecht Springer thorn Business
Media BV p 31ndash55 doi101007978-1-4020-8815-5_3
Yurkov V Hughes E 2017 Aerobic Anoxygenic Phototrophs four
decades of mystery In Hallenbeck PC editor Modern topics in the
phototrophic prokaryotes environmental and applied aspects p
193ndash214
Yutin N et al 2009 BchY-based degenerate primers target all types of
anoxygenic photosynthetic bacteria in a single PCR Appl Environ
Microbiol 75(23)7556ndash7559
Zeng Y Feng F Medova H Dean J Koblizek M 2014 Functional
type 2 photosynthetic reaction centers found in the rare bacterial phy-
lum Gemmatimonadetes Proc Natl Acad Sci USA 111(21)7795ndash7800
Zheng Q et al 2011 Diverse arrangement of photosynthetic gene
clusters in aerobic anoxygenic phototrophic bacteria PLoS One
6(9)e25050
Associate editor Esperanza Martinez-Romero
Zervas et al GBE
2908 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
- evz204-TF1
- evz204-TF2
-
Experiment grant (no 17601) and a Marie Skłodowska-Curie
AIAS-COFUND fellowship (EU-FP7 program Grant
Agreement No 609033) awarded to YZ
Author Contributions
AZ YZ LHH designed the study YZ isolated the strains
performed colony infrared imaging and extracted DNA for
Illumina sequencing YZ and AMM performed the MALDI-
TOF MS analysis AZ extracted DNA and performed ONT
Sequencing AZ carried out the bioinformatics and phyloge-
netic analyses AZ and YZ drafted the manuscript AZ
YZ AMM and LHH revised the manuscript for
intellectual content All authors read and approved the
manuscript
Literature Cited
Atamna-Ismaeel N Finkel O 2012a Bacterial anoxygenic photosynthesis
on plant leaf surfaces Environ Microbiol Rep 4(2)209ndash216
Atamna-Ismaeel N Finkel OM 2012b Microbial rhodopsins on leaf sur-
faces of terrestrial plants Environ Microbiol 14(1)140ndash146
Aziz RK et al 2008 The RAST Server rapid annotations using subsystems
technology BMC Genomics 9(1)75
Bankevich A et al 2012 SPAdes a new genome assembly algorithm and
its applications to single-cell sequencing J Comput Biol
19(5)455ndash477
Bao Z Sato Y Kubota M Ohta H 2006 Isolation and characterization of
thallium-tolerant bacteria from heavy metal-polluted river sediment
and non-polluted soils Microb Environ 21(4)251ndash260
Beatie GA 2002 Leaf surface waxes and the process of leaf colonization
by microorganisms In Lindow SE Hecht-Poinar EI Elliott VJ editors
Phyllosphere Microbiology St Paul (USA) APS Press pp 3ndash26
Beattie GA Lindow SE 1999 Bacterial colonization of leaves a spectrum
of strategies Phytopathology 89(5)353ndash359
Brettin T et al 2015 RASTtk a modular and extensible implementation of
the RAST algorithm for building custom annotation pipelines and an-
notating batches of genomes Sci Rep 58365
Carvalho SD Castillo JA 2018 Influence of light on plantndashphyllosphere
interaction Front Plant Sci 91482
Ferrera I Gasol JM Sebastian M Hojerova E Koblızek M 2011
Comparison of growth rates of aerobic anoxygenic phototrophic bac-
teria and other bacterioplankton groups in coastal Mediterranean wa-
ters Appl Environ Microbiol 77(21)7451ndash7458
Ferrera I Sanchez O Kolarova E Koblızek M Gasol JM 2017 Light
enhances the growth rates of natural populations of aerobic anoxy-
genic phototrophic bacteria ISME J 11(10)2391ndash2393
Gorlach K Shingaki R Morisaki H Hattori T 1994 Construction of eco-
collection of paddy field soil bacteria for population analysis J Gen
Appl Microbiol 40(6)509ndash517
Gourion B Rossignol M Vorholt JA Lindow SE 2006 A proteomic study
of Methylobacterium extorquens reveals a response regulator essential
for epiphytic growth wwwintegratedgenomicscom Accessed
November 26 2018
Green PN 2006 Methylobacterium In Dworkin M Falkow S Rosenberg
E Schleifer K Stackebrandt E editors The prokaryotes ndash a handbook
on the biology of bacteria New York Springer p 257ndash265
Hauruseu D Koblızek M 2012 Influence of light on carbon utilization in
aerobic anoxygenic phototrophs Appl Environ Microbiol
78(20)7414ndash7419
Katoh K Misawa K Kuma K Miyata T 2002 MAFFT a novel method for
rapid multiple sequence alignment based on fast Fourier transform
Nucleic Acids Res 30(14)3059ndash3066
Knief C et al 2012 Metaproteogenomic analysis of microbial communi-
ties in the phyllosphere and rhizosphere of rice ISME J
6(7)1378ndash1390
Knief C Frances L Vorholt JA 2010 Competitiveness of diverse methyl-
obacterium strains in the phyllosphere of arabidopsis thaliana and
identification of representative models including M extorquens
PA1 Microb Ecol 60(2)440ndash452
Kwak MJ et al 2014 Genome information of Methylobacterium oryzae a
plantndashprobiotic methylotroph in the phyllosphere PLoS One
9(9)e106704
Lindow SE Brandl MT 2003 Microbiology of the phyllosphere Appl
Environ Microbiol 69(4)1875ndash1883
Madsen AM Zervas A Tendal K Nielsen JL 2015 Microbial diversity in
bioaerosol samples causing ODTS compared to reference bioaerosol
samples as measured using Illumina sequencing and MALDI-TOF
Environ Res 140255ndash267
Martin M 2011 Cutadapt removes adapter sequences from high-
throughput sequencing reads EMBnet J 17(1)10ndash12
Marx CJ et al 2012 Complete genome sequences of six strains of the
genus Methylobacterium b J Bacteriol 194(17)4746
Noguchi T Hayashi H Shimada K Takaichi S Tasumi M 1992 In vivo
states and functions of carotenoids in an aerobic photosynthetic bac-
terium Erythrobacter longus Photosynth Res 31(1)21ndash30
Olm MR Brown CT Brooks B Banfield JF 2017 dRep a tool for fast and
accurate genomic comparisons that enables improved genome recov-
ery from metagenomes through de-replication ISME J
11(12)2864ndash2868
Overbeek R et al 2014 The SEED and the Rapid Annotation of microbial
genomes using Subsystems Technology (RAST) Nucl Acids Res
42(D1)D206ndashD214
Parks DH Imelfort M Skennerton CT Hugenholtz P Tyson GW 2015
CheckM assessing the quality of microbial genomes recovered from
isolates single cells and metagenomes Genome Res 25(7)1043ndash1055
Remus-Emsermann MNP et al 2014 Spatial distribution analyses of nat-
ural phyllosphere-colonizing bacteria on Arabidopsis thaliana revealed
by fluorescence in situ hybridization Environ Microbiol
16(7)2329ndash2340
Stamatakis A 2006 RAxML-VI-HPC maximum likelihood-based phyloge-
netic analyses with thousands of taxa and mixed models
Bioinformatics 22(21)2688ndash2690
UN 2017 World population prospects the 2017 revision Data Booklet
httpspopulationunorgwppPublicationsFilesWPP2019_DataBook-
letpdf
Vesth T Lagesen K Acar euroO Ussery D 2013 CMG-biotools a free work-
bench for basic comparative microbial genomics PLoS One
8(4)e60120
Vorholt JA 2012 Microbial life in the phyllosphere Nat Rev Microbiol
10(12)828ndash840
Wick RR Judd LM Gorrie CL Holt KE 2017 Unicycler resolving bacterial
genome assemblies from short and long sequencing reads PLoS
Comput Biol 13(6)e1005595
Wilson M Lindow SE 1994a Coexistence among epiphytic bacterial pop-
ulations mediated through nutritional resource partitioning Appl
Environ Microbiol 604468ndash77
Wilson M Lindow SE 1994b Ecological similarity and coexistence of epi-
phytic ice-nucleating (ice) pseudomonas syringae strains and a non-
ice-nucleating (ice) biological control agent Appl Environ Microbiol
60(9)3128ndash3137
Woodward FI Lomas MR 2004 Vegetation dynamics ndash simulating
responses to climatic change Biol Rev 79(3)643ndash670
Photoheterotrophs on Wheat Leaves GBE
Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019 2907
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
Xu J Gordon JI 2003 Honor thy symbionts Proc Natl Acad Sci USA
100(18)10452ndash10459
Yurkov V Csotonyi JT 2009 New light on aerobic anoxygenic photo-
trophs In Hunter NC Daldal F Thurnauer MC Beatty JT editors
The purple photosynthetic bacteria Dordrecht Springer thorn Business
Media BV p 31ndash55 doi101007978-1-4020-8815-5_3
Yurkov V Hughes E 2017 Aerobic Anoxygenic Phototrophs four
decades of mystery In Hallenbeck PC editor Modern topics in the
phototrophic prokaryotes environmental and applied aspects p
193ndash214
Yutin N et al 2009 BchY-based degenerate primers target all types of
anoxygenic photosynthetic bacteria in a single PCR Appl Environ
Microbiol 75(23)7556ndash7559
Zeng Y Feng F Medova H Dean J Koblizek M 2014 Functional
type 2 photosynthetic reaction centers found in the rare bacterial phy-
lum Gemmatimonadetes Proc Natl Acad Sci USA 111(21)7795ndash7800
Zheng Q et al 2011 Diverse arrangement of photosynthetic gene
clusters in aerobic anoxygenic phototrophic bacteria PLoS One
6(9)e25050
Associate editor Esperanza Martinez-Romero
Zervas et al GBE
2908 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
- evz204-TF1
- evz204-TF2
-
Xu J Gordon JI 2003 Honor thy symbionts Proc Natl Acad Sci USA
100(18)10452ndash10459
Yurkov V Csotonyi JT 2009 New light on aerobic anoxygenic photo-
trophs In Hunter NC Daldal F Thurnauer MC Beatty JT editors
The purple photosynthetic bacteria Dordrecht Springer thorn Business
Media BV p 31ndash55 doi101007978-1-4020-8815-5_3
Yurkov V Hughes E 2017 Aerobic Anoxygenic Phototrophs four
decades of mystery In Hallenbeck PC editor Modern topics in the
phototrophic prokaryotes environmental and applied aspects p
193ndash214
Yutin N et al 2009 BchY-based degenerate primers target all types of
anoxygenic photosynthetic bacteria in a single PCR Appl Environ
Microbiol 75(23)7556ndash7559
Zeng Y Feng F Medova H Dean J Koblizek M 2014 Functional
type 2 photosynthetic reaction centers found in the rare bacterial phy-
lum Gemmatimonadetes Proc Natl Acad Sci USA 111(21)7795ndash7800
Zheng Q et al 2011 Diverse arrangement of photosynthetic gene
clusters in aerobic anoxygenic phototrophic bacteria PLoS One
6(9)e25050
Associate editor Esperanza Martinez-Romero
Zervas et al GBE
2908 Genome Biol Evol 11(10)2895ndash2908 doi101093gbeevz204 Advance Access publication September 17 2019
Dow
nloaded from httpsacadem
icoupcomgbearticle-abstract111028955570988 by R
oyal Library Copenhagen U
niversity user on 16 January 2020
- evz204-TF1
- evz204-TF2
-